Download ppt - Things That Go Bump in the Net: Implementing and Securing a Scientific Network

Things That Go Bump in the Net: Implementing Things That Go Bump in the Net: Implementing and Securing a Scientific Networkand Securing a Scientific Network

May 15th 2013 – Merit Member ConferenceMatt Lessins – Wayne State UniversityJason Zurawski – Internet2

Outline

• Science DMZ Overview• Network Performance Expectations• Campus Security• Wayne State Science DMZ Case Study

2 – 04/24/23, © 2013 Internet2/Wayne State [email protected] & [email protected]

mailto:[email protected]


Science DMZ Overview

• Motivation– Need to move lots of data, consistently

• From scientific instrument generating data to storage & processing resources

– Physics (telescopes, colliders)– Biology (gene sequencers)

• Between different storage/processing within same research entity network

• Between different data storage/processing devices scattered around the country and the world

– Mirrored data– Distributed analysis of data






“Typical” campus network

Campus Network

Border Firewall

Border Router

Internet

Border FW and router between Internet and Campus network

Good at passing loads of smallish flows Short bursts, minimal payload Large number of flows, but

minimal impact on overall resources

FW is a roadblock in more than one sense

=





Firewalls, IDPs, Shapers and the like present “bumps” in the net

These devices can hinder the movement of large data sets (single flows)

Campus Network

Firewall

Border Router

Internet

1Gb

=





Congestion (e.g. mouse flows) is another obstacle to overcome Numerous/Short lived

flows Fills up buffers/takes away

processing time on path devices





What is needed is an End-to-End, “bump” free path

The Science DMZ is an architecture that provides a solution

Special purpose part of the network near the network perimeter

Takes its name from “DMZ networks”, a network that hosts devices outside of the secured perimeter of the organizations network





Source: http://fasterdata.es.net/science-dmz/science-dmz-architecture/



Science DMZ (In One Slide)• Consists of 3 key components, all required:

• “friction free” network path– Highly capable network devices (wire-speed, deep queues)– Virtual circuit (implementation agnostic - e.g. SDN in any flavor)

connectivity option– Security policy and enforcement specific to science workflows– Located at or near site perimeter if possible

• Dedicated, high-performance data movers– a.k.a.: Data Transfer Node (DTN)– Optimized bulk data transfer tools such as GlobusOnline/GridFTP

• Performance measurement/test node– perfSONAR

• Details at: http://fasterdata.es.net/science-dmz/


Source: B. Tierney @ ESnet

http://fasterdata.es.net/science-dmz/





Science DMZ Switch/Router Place in the Science DMZ network where security via Access

Control Lists (ACLs) takes place Attachments to Data Transfer Node, WAN, Border Router,

PerfSONAR and other Science DMZ connections that extend into the Campus network

Does it support emerging technologies such as OSCARS and OpenFlow (e.g. “Software Defined Networking” – allowing dynamic, out of band, control over network devices and protocols)?

Does it have enough buffer space to handle “fan-in” issues

Science DMZ Objects





Data Transfer Node (DTN) Usually a PC-based Linux server built out of quality components

tuned for high-speed data transfer to remote systems Could have its own local storage or be connected to SAN or

some combination Has high-speed network interfaces Runs software tools designed for high-speed data transfer like

GridFTP and versions of ssh/scp with high-performance patches Doesn’t run other software that might increase security risks, like

browsers, media players, etc… Treat this as a ‘cache’ – stage data movement through this

device

Science DMZ Objects





perfSONAR Software framework focused on federating

performance monitoring infrastructure E.g. everyone monitors internally, why not share that

information with others? Facilitates sharing of historic measurement data;

allows parties to negotiate end to end testing across multiple domains

Catches the cases that fall between the cracks Network Demarcation Application/local performance vs. true network

problems

Science DMZ Objects





WAN Not the commodity Internet, there are lots of bumps in that net National Research networks

Internet2 National Lambda Rail (NLR)

If endpoints are local you might be able to use your Regional Optical Network (RON), for example, Merit

The key is that the WAN component not be a black box No bumps Minimal number of hops

Science DMZ Objects





Campus Network

Border Firewall

Border Router

Internet

Not Internet

DTN

Science DMZ

switch/router

PerfSONAR



Outline





• The following shows how long it (should*) take to transfer 1 Terabyte of data across various speed networks**:

• 10 Mbps network :– 300 hrs (12.5 days)

• 100 Mbps network :– 30 hrs

• 1 Gbps network :– 3 hrs

• 10 Gbps network :– 20 minutes

*Can your network do this?**Assumes running at 100% efficiency, no performance problems

Reality Check: Expectations




State of the Campus


• Show of hands – is there a firewall on your campus?– Do you know who ‘owns’ it? Maintains it? Is it being maintained?– Have you ever asked for a ‘port’ to be opened? White list a host? Does

this involve an email to ‘a guy’ you happen to know?– Has it prevented you from being ‘productive’?

• In General …– Yes, they exist. – Someone owns them, and probably knows how to add rules – but the

‘maintenance’ question is harder to answer.• Like a router/switch, they need firmware updates too…

– Will it impact you – ‘it depends’. Yes, it will have an effect on your traffic at all times, but will you notice?

• Small streams (HTTP, Mail, etc.) – you won’t notice slowdowns, but you will notice blockages

• Larger streams (Data movement, Video, Audio) – you will notice slowdowns



State of Campus – Word of Caution…


• To be 100% clear – the firewall is a useful tool:– A layer or protection that is based on allowed, and disallowed, behaviors– One stop location to install instructions (vs. implementing in multiple

locations)– Very necessary for things that need ‘assurance’ (e.g. student records,

medical data, protecting the HVAC system, IP Phones, and printers from bad people, etc.)

• To be 100% clear again, the firewall delivers functionality that can be implemented in different ways:– Filtering ranges can be implemented via

ACLs– Port/Host blocking can be done on a host

by host basis– IDS tools can implement near real-time

blocking of ongoing attacks that match heuristics



State of the Campus - Clarifications


• I am not here to make you throw away the Firewall– The firewall has a role; it’s time to define what that role is, and is not– Policy may need to be altered (pull out the quill pens and parchment)– Minds may need to be changed

• I am here to make you think critically about campus security as a system. That requires:– Knowledge of the risks and mitigation strategies– Knowing what the components do, and do not do– Humans to implement and manage certain features – this may be a

shock to some (lunch is never free)



State of the Campus – End Game


• The end goal is enabling true R&E use of the network– Most research use follows the ‘Elephant’

Pattern. You can’t stop the elephant and inspect it’s hooves without causing a backup at the door to the circus tent

– Regular campus patterns are often ‘mice’, small, fast, harder to track on an individual basis (e.g. we need big traps to catch the mice that are dangerous)

– Security and performance can work well together – it requires critical thought (read that as time, people, and perhaps money)

– Easy economic observation – impacting your researchers with slower networks makes them less competitive, e.g. they are pulling in less research dollars vs. their peers



When Security and Performance Clash


• What does a firewall do?– Streams of packets enter into an ingress port – there is some buffering– Packet headers are examined. Have I seen a packet like this before?

• Yes – If I like it, let it through, if I didn’t like it, goodbye. • No - Who sent this packet? Are they allowed to send me packets? What port

did it come from, and what port does it want to go to? – Packet makes it through processing and switching fabric to some egress

port. Sent on its way to the final destination. • Where are the bottlenecks?

– Ingress buffering – can we tune this? Will it support a 10G flow, let alone multiple 10G flows?

– Processing speed – being able to verify quickly is good. Verifying slowly will make TCP sad

– Switching fabric/egress ports. Not a huge concern, but these can drop packets too

– Is the firewall instrumented to know how well it is doing? Could I ask it?



“Personal” (Software) Firewall Flow Chart




• Processing Delay: Time to process a packet• Queuing Delay: Time spent in ingress/egress queues to device• Transmission Delay: Time needed to put the packet on the wire• Propagation Delay: Time needed to travel on the wire

Causes of Jitter




When Security and Performance Clash


• Lets look at two examples, that highlight two primary network architecture use cases:– Totally protected campus, with a border firewall

• Central networking maintains the device, and protects all in/outbound traffic

• Pro: end of the line customers don’t need to worry (as much) about security

• Con: end of the line customers *must* be sent through the disruptive device

–Unprotected campus, protection is the job of network customers

• Central networking gives you a wire and wishes you best of luck

• Pro: nothing in the path to disrupt traffic, unless you put it there

• Con: Security becomes an exercise that is implemented by all end customers



Brown University Example


• Totally protected campus, with a border firewall





• Behind the firewall:





• In front of the firewall:



Brown University Example – TCP Dynamics


• Want more proof – lets look at a measurement tool through the firewall.– Measurement tools emulate a well behaved application

• ‘Outbound’, not filtered:– nuttcp -T 10 -i 1 -p 10200 bwctl.newy.net.internet2.edu– 92.3750 MB / 1.00 sec = 774.3069 Mbps 0 retrans– 111.8750 MB / 1.00 sec = 938.2879 Mbps 0 retrans– 111.8750 MB / 1.00 sec = 938.3019 Mbps 0 retrans– 111.7500 MB / 1.00 sec = 938.1606 Mbps 0 retrans– 111.8750 MB / 1.00 sec = 938.3198 Mbps 0 retrans– 111.8750 MB / 1.00 sec = 938.2653 Mbps 0 retrans– 111.8750 MB / 1.00 sec = 938.1931 Mbps 0 retrans– 111.9375 MB / 1.00 sec = 938.4808 Mbps 0 retrans– 111.6875 MB / 1.00 sec = 937.6941 Mbps 0 retrans– 111.8750 MB / 1.00 sec = 938.3610 Mbps 0 retrans

– 1107.9867 MB / 10.13 sec = 917.2914 Mbps 13 %TX 11 %RX 0 retrans 8.38 msRTT



Brown University Example – TCP Dynamics


• ‘Inbound’, filtered:– nuttcp -r -T 10 -i 1 -p 10200 bwctl.newy.net.internet2.edu– 4.5625 MB / 1.00 sec = 38.1995 Mbps 13 retrans– 4.8750 MB / 1.00 sec = 40.8956 Mbps 4 retrans– 4.8750 MB / 1.00 sec = 40.8954 Mbps 6 retrans– 6.4375 MB / 1.00 sec = 54.0024 Mbps 9 retrans– 5.7500 MB / 1.00 sec = 48.2310 Mbps 8 retrans– 5.8750 MB / 1.00 sec = 49.2880 Mbps 5 retrans– 6.3125 MB / 1.00 sec = 52.9006 Mbps 3 retrans– 5.3125 MB / 1.00 sec = 44.5653 Mbps 7 retrans– 4.3125 MB / 1.00 sec = 36.2108 Mbps 7 retrans– 5.1875 MB / 1.00 sec = 43.5186 Mbps 8 retrans

– 53.7519 MB / 10.07 sec = 44.7577 Mbps 0 %TX 1 %RX 70 retrans 8.29 msRTT



Brown University Example – TCP Plot (2nd)




Brown University Example – TCP Plot (2nd)




Brown University Example – TCP Plots






• Series of problems and solutions implemented:– 10G Firewall was not even coming close – configuration and issue with

tech support were to blame– After this, internal switching infrastructure was revealed to be

dropping packets on large flows (lack of buffering)– Mitigating step of using 1G network (not protected through firewall)

was found to be insufficient due to demand• Epilogue:

– perfSONAR Monitoring (Department and Campus) goes a long way in producing ‘proof’

– Network architectural changes to support heavy hitters will be needed– Firewalls are complex, its easy to get it ‘wrong’ in terms of

configuration. • And they need a human to watch them – its not set and forget



The Pennsylvania State University Example


• Unprotected campus, protection is the job of network customers





• Initial Report from network users: performance poor both directions– Outbound and inbound (normal issue is inbound through protection

mechanisms)• From previous diagram – CoE firewalll was tested

– Machine outside/inside of firewall. Test to point 10ms away (Internet2 Washington)

• jzurawski@ssstatecollege:~> nuttcp -T 30 -i 1 -p 5679 -P 5678 64.57.16.22• 5.8125 MB / 1.00 sec = 48.7565 Mbps 0 retrans• 6.1875 MB / 1.00 sec = 51.8886 Mbps 0 retrans• …• 6.1250 MB / 1.00 sec = 51.3957 Mbps 0 retrans• 6.1250 MB / 1.00 sec = 51.3927 Mbps 0 retrans• • 184.3515 MB / 30.17 sec = 51.2573 Mbps 0 %TX 1 %RX 0 retrans 9.85 msRTT





• Observation: net.ipv4.tcp_window_scaling did not seem to be working– 64K of buffer is default. Over a 10ms path, this means we can hope to see

only 50Mbps of throughput:– BDP (50 Mbit/sec, 10.0 ms) = 0.06 Mbyte

• Implication: something in the path was not respecting the specification in RFC 1323, and was not allowing TCP window to grow– TCP window of 64 KByte and RTT of 1.0 ms <= 500.00 Mbit/sec.– TCP window of 64 KByte and RTT of 5.0 ms <= 100.00 Mbit/sec.– TCP window of 64 KByte and RTT of 10.0 ms <= 50.00 Mbit/sec.– TCP window of 64 KByte and RTT of 50.0 ms <= 10.00 Mbit/sec.

• Reading documentation for firewall:– TCP flow sequence checking was enabled– What would happen if this was turn off (both directions?





• jzurawski@ssstatecollege:~> nuttcp -T 30 -i 1 -p 5679 -P 5678 64.57.16.22• 55.6875 MB / 1.00 sec = 467.0481 Mbps 0 retrans• 74.3750 MB / 1.00 sec = 623.5704 Mbps 0 retrans• 87.4375 MB / 1.00 sec = 733.4004 Mbps 0 retrans• …• 91.7500 MB / 1.00 sec = 770.0544 Mbps 0 retrans• 88.6875 MB / 1.00 sec = 743.5676 Mbps 28 retrans• 69.0625 MB / 1.00 sec = 578.9509 Mbps 0 retrans• • 2300.8495 MB / 30.17 sec = 639.7338 Mbps 4 %TX 17 %RX 730 retrans 9.88 msRTT





• Impacting real users:





• Series of problems and solutions implemented:– Firewall was not configured properly– Lack of additional paths to implement a true research bypass

• Epilogue:– perfSONAR Monitoring (Department and Campus) still goes a long

way in producing ‘proof’• FYI – Penn State has around 50 perfSONAR boxes now for all of their

campuses. Tremendous value from a $1,000 machine and free software– No “One Size Fits All” solution will cut it in a dynamic environment



Outline





Science DMZ (?)


• A staple of the meeting circuit for several years, Matt gave an excellent overview

• What is it really?– “Blueprint”, not a specific design– Approach to network architecture that preserves the ability to

securely manage two different worlds• Enterprise – BYOD, IP Phones,

Printers, HVAC, things you don’t know enough about to trust, and shouldn’t

• Research – Well defined access patterns, Elephant flows, (normally) individuals that can manage their destiny with regards to data protection



Science DMZ – Pro/Con on Generalities


• Pro:– Unspecified nature makes the

pattern fungible for anyone to implement

– Hits the major requirements for major science use cases

– A concept that “anyone” should be able to understand on a high level

• Con:– Unspecified nature implies you

need your own smart person to think critically, and implement it for a specific instantiation

– Those that don’t do heavy science (or don’t know they do) may feel “its not for us”

– A concept easy to treat as a ‘checkbox’ (hint: CC-NIE schools – are you stating ‘we have perfSONAR’ and moving on?)



Where the Rubber Meets the Road


• Lets start with the generic diagram (again):



Where the Rubber Meets the Road


• There are 4 areas I am going to hit on, briefly (note the last one is not ‘pictured’):

– Network Path– Adoption of “New” Technology– Security– User Outreach



Network Path


• Engineers ‘get it’– No one will dispute that protected and unprotected path will have

benefits (and certain dangers).– $, 100G isn’t cheap (10G and 40G are). You don’t have to go 100G,

implementing the architecture with existing technology is a perfectly good way forward

– You still need a security professional (if you don’t have one already) for the secured and non-secured paths. Learn to love your IDS just as much as your firewall and shapper …

• Tuning is important. Small buffers (as seen previously) make data movement sad. This means servers, and network devices

• Ounce of prevention – you need monitoring, and you certainly need training in how to use the performance tools to debug. You will be debugging (bet me a $1 if you honestly think you won’t be…)



Adoption of “New” Technology


• SDN, perfSONAR, etc. etc.– We will keep making acronyms, don’t worry

• What matters in all this? Being able to make your job easier– perfSONAR = insurance policy against risky behavior.

• Will tell you if you have done things wrong, and warn you if something breaks.

• Crucial for your campus, and costs only the price of a server, and getting an engineer up to speed on how to use it

– SDN will be a game changer. Is it ready for production (?) – hard to say. The ability to afford more control over the network to the end user relies on applications (and end users) getting caught up. Hint.

• There will be more changes in the future, it’s the nature of the game. R&E needs to be about certain risky moves away from the norm



Security


• I can spend an entire deck on this, but to keep it short:– Component based security is wrong. Needs to be a system.

• E.g. the firewall by itself has limited use, and can be easily broken by a motivated attacker

– System:• Cryptography to protect user access and data integrity• IDS to monitor before (and after) events• Host-based security is better for performance, but takes longer to implement.

Firewalls are bad on performance but easy to plot down in a network.• Let your router help you – if you know communication patterns (and know

those that should be disallowed), why not use filters?– Campus CI Plan. Make one, update it often. Shows funding bodies you

know what is going on and have plans to address risks, and foster growth• Economic argument – if you are non-competitive for grants because you

cheaped out on security, are you better in the long run?



Security - Examples


• Data Provenance– Some bureaucratic document states that all campus traffic must

be a) encrypted and b) passed through a firewall for packet inspection. Why?

• a) What data is private, and what isn’t? Student records, sure. Maybe even sensitive grant-related research. Encrypting all data is not necessary if you stop to think about the data. At least make it a user choice.

• b) Firewalls work when you can’t be sure of a traffic profile (e.g. they stop everything and give it the business). If you know the traffic profile, use that to your advantage. Data from X sites on ports Y, and Z.

– Policy is:• Written by those that often do not have practical experience• Outdated almost immediately

– Review (create) CI Plan regularly.



Security - Examples


• User Management– What is better: centrally managed user system for all

resources vs. independently managed on each machine?– Central

• Pro: Easier administration when adding/deleting• Con: Single point of failure

– Individual• Pro/Con: Breach of once machine doesn’t necessarily imply

that accounts on others are compromised (N.B. I think we are all guilty of recycling passwords though…)

– Answer depends on your campus, which is another reason why the DMZ is a blueprint, not a packaged solution



Security - Examples


• Device Profiles– All the devices are equal (untrusted)

• Have the number of phones/tablets eclipsed hard campus resources for any of you yet?

• You should absolutely not trust these, or *many* of your hard campus resources

– Some are more equal than others (trusted)• Does the Physics group have a dedicated admin who ‘gets

it’? They know Linux, and have implemented host-based security, plus split out heavy hitters from normal users?

• Give them a fast path (Penn State Model)• If policy needs to be changed, start handing out certificates

to groups that complete a training. CYA…



User Outreach


• The unstated factor:– Could you name your top 10 (5? 3?) network users? Do you know

where their traffic is going? Do you know why? Should you care?– Simple solution – (net | s)flow monitoring (pick a brand, many are

good). • Top 10 src/dst for some period of time, go and talk to the researchers.• Ask them what they are doing, how they are doing it, and if its going ok.

– Campus CI days – was a sponsored thing, but why not have one ‘just because’?

• Gets IT and research talking.• Identifies areas of growth; areas of friction

– Requires an outgoing person – hire a research engineer.• Someone who knows what a network is, and can translate statements like

“the beamline will be firing at 200Khz 2 times a week and generating 2PB of data a year” into “they need 40Gbps and a clear path to 4 international sites as well as the domestic routing table”


Science DMZ on Campus Conclusions


• A lot to consider– Security factors, when done poorly, are hurting your users in a

noticeable and significant manner– Easily found, if you have the right tools at your disposal … and you are

listening to them whine (yeah, that’s a hard one)• Its not impossible…

– Approaches like the Science DMZ are here to help– They are not turn key though

• …but it will require some thought and planning– Know your campus, know your needs– Implementation won’t take a weekend, plan for some burn in and

testing– Will pay off in the end (we promise)

Outline



Wayne State Science DMZ Case Study


NSF GrantCampus Cyberinfrastructure (CC-NIE)

Two Major components1. Creation of a dedicated high speed network to service the generators and consumers of science network traffic centered at the computing center (Science DMZ)

2. Upgrade the network infrastructure in the Physics building to modern standards



● End Points– Physics– Computer Science– Mott (Applied Genomics)– Computing Center

● Allows for -– Creation of Science DMZ– OpenFlow (software-defined networking)



Campus Network

Border Firewall

Border Router

Merit

10Gb 10Gb

Genomics (Mott)

Computer Science

Physics

perfSONAR

perfSONAR

perfSONAR

Grid Storage

LHC CMS experiment

DNA Seqencer

GENI (WiMax,ExoGENI,KanaiGENI)Bioinformatics

database

Merit/Internet2 MiLR

Fiber

perfSONAR



SDN(OpenFlow)

Low LatencyDeep Buffers

Trying to hit the bullseye



Trying to hit the bullseye • SDN

We’ll be using OpenFlow. There is an OpenFlow component in the ExoGENI project

Experiment will dictate the version of OpenFlow we’ll use What switch/router features will disappear in hybrid-mode?

Buffer size Going to worry about this at our happy confluence, the core Science

DMZ Router/Switch, not so much at the edges Rule of thumb – “ 50ms of line-rate output queue buffer, so for

for a 10G switch, there should be around 60MB of buffer” Buffer sizes not generally listed on data sheets. Will ask

around



Case Study: Mirroring a 20TB database

20TB of gene sequencing data Endpoints

Server on our HPC grid Server at UCSC



Hi,

I'm a new faculty at the CMMG that started last September. My research is in computational genomics and involves analyzing massive datasets generated by high throughput sequencing. I have been working with Michael Thomson, and I'm using the grid infrastructure for computing and data storage. I need to transfer about 20Tbytes of data from "hgdownload.cse.ucsc.edu" to my data storage server at the WSU grid "piquelab.grid.wayne.edu".The maximum transfer rate I'm getting with UCSC is about 16MBps, while from NCBI using aspera I'm getting about 80MBps. I have been working with people at UCSC to make sure it is not a protocol problem, and we think that my connection to UCSC gets routed to some bottleneck in the middle. I was wondering if the data transfer could be routed through the LambdaRail network. Do you think it may be possible to get a faster network route between UCSC and WSU?

Thank you very much for your kind consideration.

Best regards,

Roger





Decided to side-step the FW

Campus Network

Border Firewall

Border Router

Merit

10Gb

Grid Storage

Merit/Internet2



Increased throughput to 400-500Mb/s

Things That Go Bump in the Net: Implementing Things That Go Bump in the Net: Implementing and Securing a Scientific Networkand Securing a Scientific NetworkMay 15th 2013 – Merit Member ConferenceMatt Lessins – Wayne State UniversityJason Zurawski – Internet2

For more information, visit http://www.internet2.edu/research