SAP System Consolidation, Benefits and Lessons Learnedmedia.techtarget.com/searchSAP/downloads/7_30SAPFINAL.pdf · 4 SAP IT Briefing: SAP System Consolidation, Benefits and Lessons

SAP

SAP IT Briefing:

SAP System Consolidation, Benefits and Lessons LearnedBy Volker Joehnk, Platform IT Manager, HP, IT Americas

and Ken Gazarek, HP Consulting and Integration

An IT Briefing produced by

Sponsored By:

SAPSAP System Consolidation, Benefits and Lessons Learned Page 3By Volker Joehnk, Platform IT Manager, HP, IT Americas

and Ken Gazarek, HP Consulting and Integration

Table of Contents

Copyright © Hewlett-Packard Company 2003. All Rights Reserved. Reproduction, adaptation, or translation without prior written permission is prohibited,except as allowed under the copyright laws.

Design Copyright © TechTarget 2003. All Rights Reserved.

For inquiries and additional information, contact:Tina HillsDirector of Product Marketing, [email protected]

Volker Joehnk is Platform IT Manager for Hewlett Packard Company has over 17 years of SAP experience in the areas of order management,supply chain and finance. He’s held various IT positions supporting SAP applications from an operational perspective. His current positioninvolves managing day-to-day operations for HP’s largest order fulfillment SAP platform, R/3, APO and BW supporting U.S. resellers and retailers.Before joining HP, Volker was an SAP FICO consultant with PricewaterhouseCoopers and held positions as Class Controller with T-Mobile andHeidelberg Cement.

Ken Gazarek has seven years of SAP infrastructure consulting experience with HP Consulting and Integration. He’s worked as an SAPinfrastructure architect and Basis consultant. Ken has held numerous positions in SAP infrastructure design, implementation, migration andconsolidation. Prior to joining HP, Ken spent 10 years working in both IT and engineering roles in the oil and telecommunications industries.

3SAP IT Briefing: SAP System Consolidation, Benefits and Lessons Learned

Sponsored By:

JOEHNK: Hello everyone. This is Volker Joehnk andI’m with HP in Palo Alto. As Dottie mentioned, I’man IT Platform Manager supporting an SAP order-management/order-fulfillment platform, whichconsists of R/3, APO and BW. The purpose of thepresentation I’m about to share with you is toprovide you with feedback on some consolidationprojects we went through and the lessons learned.And as you can see from the agenda, which I’mabout to share with you in the next few secondshere, I would like to give a kind of overview of what

kind of IT organization we are and then continue toshare a little bit about the drivers we had forconsolidating R/3 instances, and I’m really focusedonly on R/3. I’m not talking about BW and APOconsolidations. I will move on to explaining a littlebit about types of R/3 consolidation and point youto the areas where we have done someconsolidations, and then I will very specifically talkabout two SAP consolidations we went through.Then I will wrap up the presentation with someconclusions and some lessons learned.

By Volker Joehnk, Platform IT Manager, HP, IT Americas and Ken Gazarek, HP Consulting and Integration

© 2003 TechTarget

SAP System Consolidation, Benefitsand Lessons Learned

Figure 1


Sponsored By:

Let me move on with the next slide, which talksabout our organization. We are an IT supportorganization and we report to the Imaging andPrinting Group, and as you may know, HP has fourglobal business units. I’m with the global businessunit IPG, or Imaging and Printing. The next speaker,Ken, is with HP Services, and then we have ESG,which is Enterprise Systems Group and PSG, whichis the Personnel Systems Group. We support twoSAP-based order management and order fulfillmentplatforms for the Americas, and the platform, as Imentioned earlier, consists of two R/3 instances andone BW and one APO instance. And the APO andBW instance is shared among the two R/3 instances.And we do support the replenishment business forconsumer and commercial customer accounts.

The way we are organized is simply by servicedomains. And we do have a co-sourced applicationmanagement service model in place for this venture.The organization is based on a three-tier model,which means we have help desks, partner care, andsustaining engineering —so we are not adevelopment organization. We do have respectivedevelopment organizations working on thisplatform. We, from an operational point of view, arefocused on keeping the plane flying and taking careof small stuff. Small stuff, for instance, is in effect,enhancement and system maintenance. And when itcomes to system maintenance, I’m talking aboutsoftware release levels and service packs andupgrades. And as a side note for those of you whoare interested, and I apologize if I’m getting toodetailed in advance here, we run R/3 on a 4.6cinstance release with service pack 4.0. BW wecurrently run it on 3.0B and I believe it is service pack10 and we run APO on 3.0A service pack 21 or 22.

In terms of location, I’m here at the Palo Alto site inCalifornia. We have two other sites where webasically manage the day-to-day operations for theregion here, and the region is the Americas. Theother locations we have are Toronto in Canada,basically with Accenture—who is our applicationmanagement service provider—and we do haveanother location in Guadalajara in Mexico, where webasically run the Canada and Latin Americaoperations.

Let me pause for a second and move on to the nextslide. The next slide is talking about some specificdrivers for the consolidation project we wentthrough and some of them I would say are wellknown and I think you can find them almostanywhere. When it comes to business, the business

perspective is always to lower the operational costs.What I mean by that is basically in the ordermanagement and order fulfillment business, we areexpecting to lower the cost-per-order dollars, andthis is a kind of guideline we have now going on formany years. The other aspect—and this may evenhave something to do with the merger HP andCompaq went through—is to respond in line withorganizational changes. What it means is in essenceto align the supporting IT systems with theorganizational structure.

From an IT perspective, where we are focusedheavily is, interestingly enough, on improving theservice levels. We were constantly demanded by thebusiness to increase the uptime, so we have uptimeservice levels of 98%. Interestingly enough, it wasnot good enough for the business, which isunderstandable, especially in the order managementand order fulfillment business. So, one of therequests we followed up was to increase the systemuptime to 99.5% and now we are in a higheravailability environment.

The other aspect in terms of improving service levelswas to look at failover and DRP capabilities, and Iapologize for using these acronyms. DRP stands forDisaster Recovery Plan. This is typically used orinvolved to some degree as we faced some serioushardware issues so that we can failover to backupsystems within the data center. Also, we addedcapabilities so the data center we have has thecapability of a dual site DRP where we can failoverto another site in case the entire site goes down.

From an IT operational cost structure with theconsolidation project we went through, we are nowin a position to leverage the IT stuff much better.

Figure 2

When I talk about IT stuff, it means our local stuffwe have, which is mainly focused on applicationsupport. I’m coming back to the tier-one, tier-twoand tier-three model, and of course, the IT stuff inthe data centers supporting the applications andinstances. And last, but not least, I think a veryimportant point to us is always sharing the existingenvironment infrastructure. This is one of our keypoints we took away when we went through theconsolidation project.

Let’s move on. I’m pointing to the next slide, whichgives you a kind of simplified illustration of thetypes of consolidations. The reason why I want tospend a couple minutes on it is to give you a kind ofidea what kind of consolidation techniques we have.I know that Ken, later on in his presentation, will bemore specific in some areas.

Let me start bottom-up. One of the aspects we tookinto consideration was data center migration wherewe basically looked around and said, “Okay, whatHP data center cannot provide us the service levelswe need and the operational cost structure we needin order to support the business accordingly?” Theother aspect is to look at the hardware level, and I

simplify things by saying hardware level. Mydefinition of hardware level is almost everything, soit is basically the instance, the environment, andeverything that is somehow related to hardware.The next aspect of consolidation, if you go up thestack here, is really to think about consolidationfrom client or application levels, and this would beto utilize the concept of having multiple clientswithin SAP. This means you have logical instancesand one physical instance when it comes to thedatabase, and you move your SAP application ormove our SAP application to a specific dedicatedclient.

The next level would be to consolidate within aclient. As a matter of fact, this is quite a challengeand I can talk about this later. One simplified case ofa single SAP client would be transaction processing,migration to one SAP client and this is a prettycommon thing we have utilized. I’ve seen this inother areas where you basically look at SAPinstances, and you move business from one SAPinstance to another instance. As a matter of fact, youcan do this as a business model allows if thesupporting structure is in place.


Sponsored By:

Figure 3

Let me move on with the next slide. This shares alittle bit of the experience we have made with onetype of consolidation. I think I can share it rightnow. This is referring to my earlier statement aboutmoving transaction processing from one instance toanother instance. We faced some situations aboutthree years ago that we had two SAP applicationsfor order management and order fulfillment up andrunning for the direct and indirect business. We callit direct or indirect business; it’s mainlyreplenishment business. We realized that it is prettycostly to operate and run two SAP applications inparallel, so the decision was made to retire one ofthe smaller SAP applications and to move thebusiness we had in this particular instance to abigger SAP order management and order fulfillmentinstance. So in terms of benefits, we were quiteimpressed by the savings and as you can see, werealized annual savings of about $1.4 million. Themajority of these savings are the result of somereduction of test and development systems andproduction support.

So, what it means is if you have these productionsystems up and running, you need to provide testand development systems. Plus, there isdevelopment, and yes, there are test and integration

activities going on. So, with taking away thehardware, we realized these kinds of savings rightaway, and we moved the business to the biggerinstance. For us it was more or less a sizing-upexercise in order to make sure that we could processthe additional business.

As I mentioned earlier, there were very specificsituations where you can do those kinds of things. Inthis particular case, given the fact that theseinstances are pretty similar, we only had to adddirect capabilities, mainly via consolidation, in orderto migrate the business and subsequently trade thecustomers from one SAP instance to the other SAPinstance. In order to be very specific, we stayedwithin the company code, so it was a kind ofsimplified consolidation we had at this point. Butagain, the savings we realized came mainly from the hardware.

Looking at the benefits… Let me explain the slideand clarify the purpose of it. The slide shows youthe IT operation at a glance—basically, hardwaresupport and application support. When I talk abouthardware support, I’m looking at data centeroperations and application support. This is mainlydone by us, and mentioned earlier, using the three-


Sponsored By:

Figure 4


Sponsored By:

tier model we have in place. You can see the firstline of support is the help desk, the second ispartner care and the third line of support issustaining engineering. So, going back to thisconsolidation project, we have realized savingsmainly in the hardware area. You can see welowered the operational costs significantly, and thiswas more or less focused on the hardware and datacenter operations.

We also reduced the dedicated help desk we had inplace for this particular instance and migrated thehelp desk activities to our main help desk. We wereable to reduce the headcount in order to realizeadditional savings. When it comes to partner careand sustaining engineering, in these areas wehaven’t touched anything. So we have, on purpose,not realized any savings. As a matter of fact, wetransitioned the headcount from the smaller instanceto the bigger instance when it comes to partner careand sustaining engineering.

The second example I would like to share with you,and I can spend a couple of minutes on this, is thedata center migration and hardware consolidations.As I pointed it out earlier, it’s always been a veryinteresting aspect when you look at your instance

like we did and at the data enter, to look at thecapabilities a data center can offer. I was referring tohigh availability as we looked around within HP,and we identified the data center, which basicallyaddressed our needs in terms of availability.

Secondly, we also looked at operational costs, andwe found the data center was a very competitiveoperational cost structure. So in essence, when westarted this project and successfully completed it,our operational cost structure got even loweralthough we increased the service levels. This was,for us, the underlying theme, because going back tothe business and asking for more dollars is alwaysdifficult, especially in these times. Somehow weresolved the issue by identifying a data center andincreasing the service levels and lowering theoperational costs.

As a side note, we also realized savings, because weran our entire platform on an HP 9000 machine.With moving to this data center, we were able totake advantage of the Superdome architecture. Thearchitecture is a very compelling architecturebecause of two reasons. You can basically stickeverything into a couple of Superdomes, and yourun multiple instances like BW and APO and R/3 in

Figure 5

one Superdome. You share the disk storage, so thereare some very compelling reasons to utilize theSuperdome architecture. Furthermore, you can savea lot of floor space. So, if I look back at the HP 9000architecture, I had used up a lot of floor space, andwith the Superdome architecture you can basicallygain a lot of floor space. This is an important pointwhen it comes to high availability data center,because I don’t want to go ahead and flood the datacenters with servers in the subject location.

Coming back to the data center migration andhardware consolidations, the situation I’m referringto is, again, that we support two SAP applications;R/3 applications for the order management andorder fulfillment space. These two applications arelocated at two different data centers and supportedby two different IT operations staff when it comes tothe data center. So, we went ahead and launched theproject about six months ago, where we simplystarted the consolidation project of moving theapplications and the entire instance, we have threeinstances, from one data center to the other datacenter. In addition to that, we took advantage of theSuperdome technology we had in place and addedcapacity to the existing Superdomes in order toleverage the hardware much better.

So in essence, if you look at the benefits, you can seetwo key highlights. First of all, the service levelshave improved. The system uptime went up by0.5%. We are right now on 99.5%. We have addedthe failover and DRP capabilities, and we are in aposition right now of sharing the infrastructure andenvironment much better. We couldn’t do it beforebecause we were basically located in two differentdata centers. Even on top of this, we were able tolower the operational costs. As you can see from thelast statement, we realized an ROI in less than sevenand a half months. What it means is that for us, itwas a very successful project because these ROIs, ifyou can do it in less than one year, it’s excellent. Asyou can see even from the cost reductions, werealized savings about 35% with the combination ofmoving into a data center, increasing the servicelevels, and sharing, consolidating and leveragingexisting hardware we had in place.

Moving on to the next slide… Let me show the grida little bit. Again, as I pointed out, we havesignificantly improved the service levels. Let mestress again, especially when you are in the ordermanagement/order fulfillment business, availabilityis everything. Even if we face hardware issues, weare in a position now with the failover capabilities


Sponsored By:

Figure 6

Sponsored By:


Figure 7

Figure 8

that we have to failover to standby servers withoutinterrupting the business. I think this is the mostimportant benefit and, from my perspective withmanaging the day-to-day operations, if you havehardware issues like we had in the past, we cansimply failover and we can continue the operationswithout any business impact.

The other big saving we have realized, and this iscoming back to the statements I made earlier, is wehave reduced the hardware demands significantlyby utilizing the existing hardware we have in thedata center. This is, again, a Superdome-basedhardware and architecture we have utilized.We havetaken advantage of the operational cost structure inthis data center, which was more in our favor thanthe cost structure in the other data centers we haveexperienced.

Coming to the conclusion, of these consolidationprojects we went through, brings me to the last slide.Again, one of the key drivers for SAPconsolidation—and this is very specific to thebusiness we are in—is lowering the operational costsby increasing the service levels. As I stated earlier,we were very successful in doing this, although it ishard work, there is no doubt about this and there’s

no easy going in terms of achieving these goals. Wespent a lot of time in standardizing and simplifyingthe services among the order management andfulfillment instances, and this is mainly to improvethe productivity and efficiency we have in this area.

The second finding, or conclusion, we took awayfrom these consolidation projects is the data centerlocations and the hardware locations—and I think Istressed the points multiple times—the location ofthe data center is very important to us for tworeasons. These are labor costs plus capabilities thedata center has to offer. Especially in our business,where we have to provide high availability to thebusiness, we found the data centers within HP—which had the high availability capabilities, failovercapabilities, plus dual site disaster recovery plan—useful. On top of that, we were able to utilize thehardware much better. Looking at the hardwareitself, you can utilize the application and databaseservice up to 50 – 60%, which is a pretty goodnumber for us. If you’re interested in response time,at one point we were able to achieve response timeson the R/3 system of less than 200 milliseconds,which is pretty good and almost far too good forthese kind of instances.


Sponsored By:

Figure 9

Then the other take-away from these consolidationprojects is we typically break these projects intosmall pieces with a fixed scope and timeline. Giventhe fact that these projects are risky, especially whenyou move things around, we felt that breaking theseprojects into smaller pieces gave us the opportunityto be successful. We typically do not allow anychanges to the scope and to the timeline because,especially in the business we are in, once we getagreement from the business to a migration scheduleand to a fixed schedule, we need to stick to it andwe cannot push it out. So, one of the key drivers forus is to break these projects and programs into smallpieces.

This is not the end of the story. As a matter of fact,we will continue to go into the next consolidationphases with our respective development teams.Those consolidation phases will be more focused onthe client level where we then go ahead and say,“Okay, let’s think about multiple R/3 lines,company codes running on one instance, on onephysical instance,” and these are things we are nowapproaching and one or two of the investigations wehave underway. Hopefully in the next three or fourmonths, we will be in a good position to share the

plan and how we want to move on with the clientconsolidation.

That’s the presentation I have in terms of ourexperience with the consolidation projects, and Iwould like to turn it over to Ken.

GAZAREK: Thanks, Volker.

I’m Ken Gazarek, and I’m with HP Services,Consulting and Integration. My role in the last sevenyears has been to work directly with customersimplementing SAP infrastructures, migrating SAPfrom one platform to another or doing consolidationswhere we were doing full technology refreshes,consolidating instances or just streamlining andoptimizing the SAP infrastructure environment.

Today I’d like to give you an overview of variousconsolidation topics. Consolidation is such a topicthat it would take hours to discuss the many, manypoints involved, but I’m going to try to survey allthe different points so that we can have a goodoverview of everything involved in an SAP systemconsolidation.


Sponsored By:

Figure 10

To start with, I want to talk about the SAPconsolidation roadmap and what a typical journeywould look like for a company as they would movefrom how they were implemented five to six yearsago when they first deployed various applications toa highly optimized, almost utility model wherethings are set up like the phone company. There arevarious stopping points and paths along the path,and this is just a model for a journey. For everycompany, this journey is very different. Most of thetime, it’s not a big bang approach because obviously,there are tons of risks and investment involved here,but in general this is the path that companies wouldtake.

Starting out where systems are physically andlogically distributed, when you first did yourimplementation, your division had a budget, theyhad a project; you implemented a product andstarting running with that. Then another divisiondid the same thing, and so on and so on. You woundup with all these disparate systems within thecompany.

The first step to consolidating SAP and optimizinginfrastructure is the co-location phase. Really, thefocus here is to try to take all of these disparatesystems and try to at least put them in the samelocation: a single, secure, managed data center. Tryto implement management systems, like OpenViewfor instance, that gives you visibility into each ofthese systems: what’s going on, which ones aredown, which ones are up, which ones areperforming well and which ones are performingpoorly. That’s really the first step in helping to get ahandle on and manage all of these systems.The second step is more of what we frequently thinkof when we think consolidation. That’sconsolidating hardware and integrating varioussystems together. What we’re looking at here isimplementing things like storage area networks andlooking at putting partitioning in systems where wecan put multiple applications together on a singlebox and reduce the amount of maintenance involvedwith the SAP systems.

The next box down the line, the applicationintegration place, is a place where very fewcompanies have gone. It requires a significantinvestment and involves the actual consolidation ofthe applications themselves. Where you see thismost frequently is when two companies mergetogether like with the recent merger of Compaq andHP. We’re merging our HR systems and payroll andreally couldn’t operate with two completely separate

sets of systems, so we had to make the investment,do the business process re-engineering and geteverybody onto a single set of application systems.But, this is very time-consuming and requires asignificant investment.

The final step in the journey is kind of a vision totreat IT almost like utilities—like the powercompany or the telephone company, where you’vegot a highly standardized set of servers. Think ofmaybe Blade servers, et cetera, and then somesoftware that, when you have a new application,you can carve out x amount of horsepower, xamount of storage. The software then automaticallycarves up your resources for you and makes themavailable to your end users for their applications.That’s a pretty long path from the most disparatesystems to running like the utility data center. We’vegot some tools and techniques that can get you tothat process along the way.


Sponsored By:

Figure 11

Figure 12

The next thing I’m going to talk about here is thevarious SAP consolidation techniques, and we cantalk about the different ways that you canconsolidate SAP.

To start with, there’s a term called applicationconsolidation. This is what a lot of companies havedone over about the past eight years or so wherethey’ve taken a lot of... I think the whole year 2000was probably the starting point of this withcompanies trying to get away from all theirhomegrown systems that had huge investments incustom code. They were becoming very un-maintainable and had this issue that all this had tobe year 2000-compliant. A lot of companies decidedat that point in time to look for a standard softwareoffering such as SAP and consolidate on a number ofapplications that they ran in their environment, andtried to get off their custom applications as much aspossible. This is what a number of companies havedone already: moving from their custom apps or anumber of apps from different vendors andstandardizing on a single, consolidated applicationlike SAP.

The second point down the line is SAP server

consolidation. That’s going to involve consolidatingapplication servers and lower horse-powered boxeswith some of the more recent technology—fasterservers with much more bandwidth, much moreCPU processing power—and reducing the overallserver count in the environment. Still, the serversremain in their own environments, but takeadvantage of some of the new, latest stuff availableon the market.

The third item down the line, SAP systemconsolidation has to do with taking multiple SAPsystems and running them on single servers andconsolidating. There are a couple of ways to do that.One technique is called Application Stacking, whereyou would run multiple SAP systems within a singleoperating system instance. Things like developmentor standby systems are typically done this way,when you can afford to have many different SAPapplications running on a single operating system,and it’s without the worry in a firm’s liability andstability, as you would in a production environment.Also, SAP for smaller companies has the capabilityof running multiple components in a singledatabase, in what they call MCOD where you canhave R/3 and BW and CRM all running on a single


Sponsored By:

Figure 13

Oracle database. We try to reduce the amount ofadministration costs and hardware costs involved inthat environment.

The next item, number four down the line, SAPinstance consolidation is very similar to what Idescribed before when we had the two companiescome together and we had to merge the actual SAPsystems themselves. So, when Compaq was runningtheir client and HP was running their client, we hadto merge those two clients together so everybodywould be running on a single instance. Thatrequired a significant amount of business process re-engineering and application development.

The fifth item down the line, SAP storageconsolidation, which I’ll get to in some of my furtherslides, is basically talking about taking all of thededicated storage for each application that wementioned in the past, maybe it was SCSI attached,maybe it was what we used to call JBOD (just abunch of disks), directly inside the server itself andconsolidating them into a storage area network andshared storage pool that could be used much moreefficiently and could improve the performance of thesystems.

Number six is consolidation and partitioning. This isone of the biggest opportunities these days toconsolidate multiple applications within a singleoperating system instance or multiple operatingsystems instances on a single box and see someefficiencies there from an operations cost, licensingcost, et cetera.

Then finally, number seven is consolidation andavailability. Before, we might have either no highavailability solution or dedicated high availabilitysolutions where you’d have a dedicated failoverserver for each production box to maybe one tomany failover scenarios. We could have manyproduction boxes potentially failing over to a singlefailover box, or you might have an SAP productiondatabase failing over to a production applicationserver. Now, maybe you look at failing it over to aquality assurance server or something, so you’re not taking out production capacity when you have a failover.

The next slide has to do with HP’s consolidationmethodology. As it mentions our approach atlooking at one of these consolidations, we’re tryingto make the IT environment more efficient within a


Sponsored By:

Figure 14

company. As Volker mentioned, this is notsomething that you want to do using a big bangapproach. This is something that you look at insmall pieces trying to balance the risk and theinvestment and trying to get the payback as quicklyas possible from that investment.

The approach we would typically take is to start bydoing some discovery work early on and thenholding something called the value andconsolidation workshop where we would bringtogether all the key stakeholders in theorganization—your IT managers, the end users, etcetera. There we discuss and agree on objectives andtry to come up with what the success factors are andthe time frames for each of those objectives inconsolidation. Then, we try to come up withopportunities to meet those objectives and come upwith timeframes and independencies between thoseso that you’ll have some ideas on how to moveforward. Then, we try to take all those opportunitiesand come up with a series of smaller projects thatwe can get that quick return on investment andminimize our risks. Finally, before we close theworkshop, we try to get an agreement from all thekey stakeholders and players that are there on that

list of projects so that we can go forward with thenext step which is to present it to executives withinthe company. If they’re buying into it, then we moveforward with two tandem processes, one of which isdoing the high-level design for those projects, andalong with that, doing an investment justification,kind of as a total cost of ownership analysis on thoseand trying to determine the ROI.

Then, we review it again with the executives andmove into a detailed design phase where we look atthe various tools for implementing each of thoseproject plans, et cetera. Then, we have anotherreview with the executives and then go into theimplementation. This is an ongoing process afterwe’ve had our first number of successes, then wekind of go back in and continually move forwardand refine the environments.

The next slide talks about various solutions availableto help us on those consolidations. They vary quite abit, from having your systems as isolated as possibleto sharing a single operating system instance. All theway on the left, we’ve got a concept called thehyperplex which allows us to combine multiplesystems, multiple operating systems with an


Sponsored By:

Figure 15

incredibly fast backend to link everything together.

Moving to the right we’ve got nPartitions, which areknown as hard partitions within a number of ourboxes including our Superdomes. Here we canassign a number of CPUs, specific amounts ofmemory, specific boards to each operating systeminstance and we can have a number of productionsystems physically isolated from each other where acomplete partition could go down, a hardware issueor whatever else, and not affect the other partitions.But yet, being on a single physical box like theSuperdome, that gives you a lot of flexibility tomove things around.

Here we move on to virtual partitions, which givesyou complete software isolation but gives you theability to share your resources amongst your variouspartitions. You can basically assign x amount ofCPU to one partition, x amount of CPU to anotherpartition, x amount of memory to one partition,memory to another partition but yet have the abilityto move that borderline and move resources fromone partition to another more flexibly than with thehard partitions.

Then finally on the right, we have resourcepartitions commonly known as application stackingwhere we stack multiple applications all within thesame operating system instance. This is somethingthat you might have typically done in your sandboxdevelopment environment before. We’ve got a lotmore flexibility now to manage resources withinthere. Before we might have, let’s say, two sandboxinstances running an operating instance and if oneof them had a runaway process, it would chew upall the CPU in there. Or if it had a memory leak, itwould chew up all the memory in the process therein that particular operating system instance. Nowwe’ve got some tools like Process Resource Managerto allocate a minimum amount of CPU to particularprocesses or Workload Manager saying, well, wewant to guarantee x amount of response time for aparticular process and try to manage the resourceson the box to more equitably distribute them amongthe applications that are residing in that singleoperating system image.

The next slide talks about a similar tool for theWintel platform, VMware and Vmware, just like thenPars and dPars in Unix allows multiple operatingsystem images to run on the Intel platform. Things


Sponsored By:

Figure 16

like Web servers are perfect candidates for whereinstead of having a dedicated server for each ofthese Web servers, you could create individualoperating system partitions and have them run on asingle box.

The next slide talks about storage consolidation andthe use of storage area networks. With storage areanetworks, each system will have its own dedicatedstorage. Because each system didn’t require a ton ofstorage, you might find that it wasn’t used quite asefficiently as it could with the storage area networks.You might have some extra space on one system thatyou couldn’t allocate to another system, and youmight have everything allocated and assigned toonly maybe a few platters instead of being able tospread it out across a ton of platters and the storagearea networks. For instance, if application one hadan immediate need for storage, you could take itfrom some unused storage that was assigned tosystem two and move things around much, muchmore flexibly.

The next slide talks about another dimension to that,where we look at not only consolidating a SAN, butalso consolidating the different applications andputting them on the same set of physical disks. Oneof the latest best practices for disk layouts is stripingnear everything. It is really fantastic. About six yearsago, one of our performance folks came to ourimplementation group and said, striping neareverything. We said, “Are you crazy? SAP doesn’trecommend that. They say put your reading logsover here, put your copies of your reading logs overhere, put your SAP tables over here, put yourindexes over here and keep everything separate.”But after many years and many discussions thatwe’ve had, we’ve really learned that consolidatingand striping near everything is the best practice.We’ve seen huge response time improvements bydoing this. There are a lot of tricks to this, forinstance like the production systems — keepingthem on the outermost cylinders of the platters forbetter IO throughput, systems with less throughputrequirements coming on innermost cylinders andkind of spreading things out. Spreading thingsacross many more physical platters, we’ve seensome huge improvement in response time.

The other thing that we’ve been implementing quitea bit over the past few years is tape area networks.So this is a different twist on the storage areanetworks for tape. Before, you might have adedicated number of drives — either, I guess DDFsor DLTs or SuperDLT directly connected with the


Sponsored By:

Figure 17

Figure 18

Figure 19

SCSI connections to each of your servers forbackups. Now through some of the new technologythat’s available with FibreChannel and fibreswitching, we can go ahead and free the network foreach of the tape drives, and a large tape librarywould look like it exists on every single box on thatstorage area network and through the use ofintelligent software like HP Data Protector, andmanage that environment. You could have everyserver on your storage area network have potentialaccess to any of the tape drives in your tapelibraries. There’s a lot of flexibility and much betterreturn on your investment in your tape hardware.

The other technology that we’ve been using quite abit over the past many years is the zero downtimebackup capability. This exists in many of thedifferent vendor solutions out there whereessentially you take your SAP database on theapplication server. I guess application server is a badterm; it should really be database server. Take allyour table spaces, put them in backup mode, makethem consistent and then on the disk array itself hasan extra mirror of disks. The top form that’s labeled“Apps” would be your primary disk for yourdatabase, which would be a normal mirror SAP.

But then, you have a third mirror, that number Tdown in the bottom that, once the table space is allput in backup, you’d split that up from the primary.Take all the table spaces out of backup mode, thenmount that T copy of the database onto the backupserver and then back it up to the tape library. One ofthe challenges lately is that more and more of ourcompanies have been going global but trying to finda good window to do backups. Our users arestarting to use them more and more, and as it’sbecome more and more global, our peak times aredistributed and spread out more and more. We’vegot a challenge just to try to find windows to dobatch activity. To try to do backups on top of thattends to be very difficult. So, by offloading that extrabackup port through a backup server, we’ve beenable to manage some of our resources much betterand minimize the impact on the applications or thedatabase servers themselves from doing backups.

The next slide talks about high availability and someof the things we could do in that space as we look atconsolidating. Before, where we might have had adedicated server for failover or we failover to an appserver, we now can failover, through the use of SANtechnology, multiple applications like R/3, APO,BW, all to the same failover box. Because we were onthe SAN, we could map those lines out to any of theboxes that are attached to it.

Another thing we’ve seen quite a bit lately isoutsourcing and moving the hosting of the systemsout to a third party provider. HP Managed Servicesprovides a number of different outsourcingopportunities, either managing hardware onsite orremotely at one of their facilities. They can provideany level of system support from peer, light power,et cetera by Unix or Wintel system administrationthrough database administration or SAP Basis. Thereare many different opportunities in a very flexibleenvironment. They can customize SLAs to meetyour specific business needs — very flexible on theinvestment themselves.

Either you can continue to own your assets, or HPcan own the assets or a number of different leasingoptions are there. There are a ton of customers thatare on different HP Managed Services data centers.A couple of them are within fairly close proximity toeach other: far enough that if a disaster happened atone, the other would be protected, yet close enough


Sponsored By:

Figure 20

Figure 21

Sponsored By:


Figure 22

Figure 23

that we could take advantage of some of the disasterrecovery tools like the business continuitytechniques, like continuous access XP or we couldmirror the primary system to another disk frame atthe other site. If one of the sites went away for somereason, we could then start running the businessfrom the other site. All of these sites are set up withredundant power, redundant telco, and redundantconnections to the servers and half from one side ofthe row, half from the other side of the row. It’s beenvery impressive to look at these sites. We justfinished a project that’s at one of the data centers upin Toronto and I was impressed with what I sawthere. A number of these data centers were createdfor the banking and finance industry, which requiresan incredible amount of high availability, andthey’re probably the best data centers ever seen.

Now, HP Services also brings a number of differentthings to the table that can help you with yourconsolidations. We’ve got very deep SAP and Oracleknowledge within the consulting practice: certifiedOracle DBAs, certified Basis consultants, and veryclose relationships with the main vendors like SAP,Microsoft and Oracle. We’ve got the experienceworking over many years with different customers

with the best practices that we could bring to thetable and how to best approach a consolidationproject, how to optimize your IT environment anddeploy solutions and also transfer that knowledge toyour folks so that you can continue moving downthat direction after our first few projects arecompleted.

Here’s just a few of the different solutions that HPServices can bring to the table, from IT infrastructuredeployments to upgrades, migrations, consolidations,capacity planning, performance tuning...a wholenumber of different things that we could assist yourcompany with.

Now HP also has some unique technology solutions.Looking at the compatibility, being able to runmultiple applications on a single box like theSuperdome, where we can run Linux and Windowsand HP/UX wrapped all in the same box. Then,remote access and our remote console access as welook at consolidating the servers to central locations.They might not necessarily be at the same place thesystem administrators are at and that’s one thingthat’s helped us quite a bit. Remotely managing theservers from a console perspective, remotely


Sponsored By:

Figure 24

Sponsored By:


Figure 25

Figure 26

managing the tape devices, storage, et cetera.worked out very nicely. We also offer greatmonitoring utilities like OpenView, resourcemanagement and storage management utilities.

Finally, HP also has unique processes that can helpwith consolidation and optimizing yourenvironment, like high availability, trying tooptimize your uptime as much as possible and thereare all sorts of redundancy solutions that can beimplemented helping with SLA management.Looking at things like HP Service Navigator, whereyou can look at all your boxes from the service orbusiness perspective and see at any one point intime what the status of that is, and then alsobringing to the table various consultingmethodologies, outsourcing opportunities or variousfinancing options for hardware and software and/orconsultant services.

All right, that wraps up my set of slides and now I’llturn it back over to Dottie for questions.

We are now moving on to the live Q&A with youour audience. Please feel free to enter any questionsyou may have at this time by clicking on the “Ask a

Question” button in the lower left corner of yourpresentation screen.

I do want to give Volker and Ken just a moment tostart reviewing your questions that have alreadystarted to come in while I give another quick thanksto today’s webcast sponsor, HP. This webcast ispresented to you today by HP. And as a special offerto today’s webcast participants, HP is makingavailable a number of white papers from Gartnerand IDC to help you plan for and implement yourown SAP consolidation. Also for a limited time, savean extra 20% on HP servers when you upgrade SAPR/3 and purchase a qualifying server. Click on thebanner ad in the lower right corner of yourpresentation screen to download the white papersand get full details on the discount promotion.

Okay, Volker and Ken, back to you for the live Q&Asession.


Sponsored By:

Figure 27

Question & Answer Session

with Volker Joehnk and Ken Gazarek

JOEHNK: Thank you, Dottie, and I see a couple ofquestions coming in so let me start off with the firstquestion from John Flynn. Ken, help me out anddive in if you have any additional comments.

John asked about the human and organizationalchallenges we have encountered with theconsolidation.

A. And yes, we’ve had some challenges. Notnecessarily resistance to change, but there were a lotof concerns about that and the way we addressedthis typically is through communication among andwithin the teams. When it comes to the business, wehave a model in place where we have dedicatedoperational IT account managers who are ourinterface to the business. So if I’m the business and ifI’m concerned, I typically go to my dedicated Opsaccount IT manager and I will share my concerns,and the Ops IT account manager will take care of it.So from a communications point of view, we havedone these kinds of measures in order to minimizeany risks and challenges we have with theseconsolidations.

And the questions about forced moves? Yes, the –sometimes the work shifts of individuals but weaddress this, but layoffs… Typically we don’t haveany layoffs in these kinds of scenarios.

Are there any other comments from yourperspective, Ken?

A. No. I think that’s a pretty good summary.

And, John, if you want to get more details, you’refree to touch base with Ken or with Jesse and youwill see one slide will show the contact names on theHP side.

Q. Another good question is, and this is comingfrom Jackie, she asks us about how do we handlethe missing windows for maintenance?

A. This topic would last, I would say, a minimumfor a two-hour session, Jackie. Let me summarizethis as quickly as possible. We typically have aplanned downtime window of eight hours persystem a month. So the overall downtime we haveper year is 96 hours. Within the eight-hourdowntime window, six hours are reserved for

maintenance and two hours are reserved for movingto production, so we have developed sometechniques, really, to shrink down the maintenanceand the move to production windows, as we call it.To give you a kind of idea of when it comes tomaintenance windows, typically what we do is weuse online reorg tools in order to reorg the tables sowe don’t have any offline reorgs, because we cannotafford this because offline reorgs would mean takingthe system down.

So if you have more specific questions about thetechniques, again, I would suggest that you touchbase with Jesse or some of the HP folks and we cancertainly assist and help you in terms of giving youa kind of ideas on how we have optimized andshrinked these downtime windows.

Another — Yes, go ahead.

Q. Next there’s a question from Houston, Sara:what was the impact of system performance andwhat do you feel were the factors that influencedthe impact?

A. Looking at system performance, we just finisheda project up north where we consolidatedproduction environments about ten productionservers down to two. Because of the advances oftechnology and CPU horse power, SAN, and IOthroughput that we’ve gained by distributing disksacross many more spindles and distributing thatwith the other systems, we were able to see hugeperformance gains. When we first approached thisdesign with the customer, they couldn’t believe us.You can’t take that many boxes, that many CPUsand consolidate the two different operating systeminstances and see that much performanceimprovement, but we did. You know, we went andwe saw better than a full tenth in response timeimprovements from this recent technology refreshand consolidation.

Q. All right, another question came in from SteveFulder about what are the key issues with runningmultiple lines on a single hardware instance?

A. One of the key issues you’re facing is themaintenance window because you need tocoordinate those maintenance windows among theclients. So what it means is: if you have hardware


Sponsored By:

maintenance to perform, you need to take down theentire system. So multiple clients are impacted. If themultiple clients are on different regions, then youneed to coordinate the downtime windows, which isa challenge. As a matter of fact, we experiencedthese kind of things and it took us quite some timeto get these issues addressed, but this is for me, Iwould say, the key issue if you utilize the multipleclient concept for — between regions.

Q. Another question came in. How to migrate ourthree BW, CRM, and APO into one system usingMCOD?

A. That’s a question I cannot answer easily. I wouldsay, touch base with Ken or with Jesse to give youmore details on what HP can offer. We, from IPG ITAmericas, have basically no experience with thiskind of feature.

There’s a real good OSS note that describes theprocess or if you contact our organization, we’d beglad to walk you through that as well.

Q. Another question came in from Steve Johnsonasking that they have four SAP landscapes on a4.0b client, wants to consolidate to one SAP withbusiness data configuration upgrade to 4.6x withhaving all data. Is this possible?

A. I don’t want to get philosophical. Everything ispossible. I think this is a question that really needsto be investigated in what kind of options andopportunities you have. I’ve seen similar questionsbefore saying, should we go from upgrade to fullconsolidating things or should we go andconsolidate first before we go to an upgrade? It verymuch depends on the situation you are in. To giveyou one example, it depends how clean your SAPinstance is in terms of customization. There are a lotof things that need to be factored in and I think Imentioned this in another session, but I typically donot see any kind of cookie-cutter concept where yousay you need to do this in order to achieve certainresults. It very much depends on the situation youface on the health of your systems and theenvironment you have and the business you support

Q. Another question came in about multipleclients’ production system managing clientindependent changes.

A. Yes. Big challenge, especially when we touchclient-independent tables. As a matter of fact, from a

development perspective, we have a process in placewhere we basically manage those kinds of changerequests so that we typically do not override orchange existing entries. I would say it is more or lessan organizational issue rather than a technical issueand it can be controlled by a process and by limiting,I call it now, a user access or developer access whenit comes to profile changes. In other areas we havetypically one or two key people who are overseeingthose kinds of changes, and they are basicallyempowered to approve those kinds of changes andbe the point of contact. So, again, it is more or lessfrom a process organizational perspective where wehave addressed those kinds of things.

JOEHNK: There are a lot of questions about gettinga soft copy of the presentation so I don’t know atthis point how we want to proceed on this.

I think, again, if there are any questions, I wouldpoint it to Jesse and Ken or someone should sharethe last slide outlining the contacts from the HP site.

GAZAREK: In terms of a soft copy, there’s actually awhite paper that’s going to be produced that will beonline on the SearchSAP.com site, and it usuallytakes about four to six weeks to take the transcriptand put it all into a white paper format. But it willbe available. If you need information ahead of time,feel free to use any of the contacts that are up on thescreen now.

JOEHNK: One question was about the multi-vendordata center and let me elaborate a little bit on thisand Ken pointed to a data center we, HP, happen torun for them. As a matter of fact, it’s a multi-vendordata center, so HP supports even on HP hardwareand equipment in this particular data center, so thereare a bunch of opportunities even if you are on non-HP equipment to consider those kinds ofopportunities. As I mentioned when I worked in thisdata center I’ve seen at least products from five orsix competitors running under this particular datacenter and it’s a very, very impressive data center.As Ken pointed out, the data center primarilysupports the banking and finance industry, whichhas very specific requirements when it comes toavailability and DRP and failover.

All right. I think that’s, in terms of questions, Ken ifyou see any other questions you want to answer?

GAZAREK: No, I think that’s about all of them.


Question & Answer Session with Volker Joehnk and Ken Gazarek

Sponsored By:

MODERATOR: Okay, great. Thanks so much,Volker and Ken.

Well, folks, that concludes today’s webcast, “SAPSystem Consolidation: Benefits and LessonsLearned.”

If you’d like to review today’s material at some laterdate, an archived version of this webcast will bemade available on SearchSAP.com’s online webcastpage within the next 24 hours.

I would like to give a final thinks to our guestspeakers, Volker Joehnk and Ken Gazarek for theirtime, knowledge and expertise today. We reallyappreciate you joining us. A very special thanks toour sponsor, HP, for bringing us this presentation.I thank you all so much for joining us today and Iwish you all a good day.


Question & Answer Session with Volker Joehnk and Ken Gazarek

Sponsored By:

About TechTargetWe deliver the information IT pros need to be successful.

TechTarget publishes targeted media that address your need for information and resources. Our network of industry-specific Websites gives enterprise IT professionals access to experts and peers, original content and links to relevant information from acrossthe Internet. Our conferences give you access to vendor-neutral, expert commentary and advice on the issues and challenges youface daily. Practical technical advice and expert insights are distributed via more than 100 specialized e-mail newsletters, and ourWebcasts allow IT pros to ask questions of technical experts in real time.

What makes us uniqueTechTarget is squarely focused on the enterprise IT space. Our team of editors and network of industry experts provide the richest,most relevant content to IT professionals. We leverage the immediacy of the Web, the networking and face-to-face opportunities ofconferences, the expert interaction of Webcasts and Web radio, the laser-targeting of e-mail newsletters and the richness and depthof our print media to create compelling and actionable information for enterprise IT professionals. For more information, visitwww.techtarget.com.

About SearchSAP.comSearchSAP.com provides SAP-specific search results from a comprehensive SAP-based domain, collects and reports the mostnoteworthy SAP news, finds the best SAP-related links on the Web and provides technical tips for SAP managers, administratorsand developers. SearchSAP.com is a key player in the SAP community with nearly two dozen industry experts on its Ask theExpert panel. Through interactive Webcasts and Ask the Expert, SearchSAP.com provides free technical advice and assistance tomore than 100,000 SAP users worldwide. www.sap.com

Documents

SAP System Consolidation, Benefits and Lessons Learnedmedia.techtarget.com/searchSAP/downloads/7_30SAPFINAL.pdf · 4 SAP IT Briefing: SAP System Consolidation, Benefits and Lessons