HP Customer case study: considerations for choosing a converged infrastructure vendor

TAKING BETTER CARE OF PEOPLE

OUR OBJECTIVE

HAS ALWAYS BEEN

92+ YEARS, TAKING CARE OF HEALTH IN BRAZIL …

358 beds

4,700 employees 3,800

doctors

250,000 laboratory exams

per month

25,000 imaging exams

per month

3,4 million patients

ICU area

2,600m² & 40 beds

1,200 surgeries per

month

19 Operating rooms

*2012 numbers

1970

1921

2014 1990

TAKING BETTER CARE OF MORE PEOPLE

2016

40,000

2016

32

Operating Rooms

2016

30,000

SURGERIES

2016

710 2012

358

Beds PATIENTS

20,000 2012

TAKING BETTER CARE OF MORE PEOPLE

PACS

VOIP

Video

Network Data

Nurse Call

Video Surveillance

Building Automation

Hospital Information

System

Infrastructure of Technology Information, why so important ?

© Copyright 2013 Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice. 7

VRRP+RSTP/MSTP between switches • Complex network design – PVSTP+ • Redundant links, complex routing planning • Long recovering time • Node and link failure cause instability • Any link failure will cause topology change

• Regular Link Aggregation between Cores • No chassis syncronization/redundancy • Proprietary routing protocol – EIGRP • ACTIVE / STANDBY Solution • Everything collapse in the Core (Servers, Agg, etc)

HSL Previous Network Topology Overview

C6500

C2960S

C2960S

C2960S

C6500

Servers


HSL New HP Network Topology

EVA Storage C7000 Servers HP12518 HP12518

HP 5900 HP 5900 HP 5900 HP 5900

HP 7510 HP 7510

Datacenter Layer

Distribution Layer

© Copyright 2012 Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice.

HSL Network Datacenter Solution


12500 series

HP Data Center Core Switching

• High-performance 1G/10GbE with future 40G/100G connectivity for DC Core/EoR

• HP IRF technology delivers resilient virtual switching fabric

• Highly available design with hot-swappable fans/power/modules/fabric

• Robust software capabilities with Full L2/L3 feature set including MPLS

• Reduced power consumption

HP 12508 HP 12518 HP 12504

13.32Tbps switching capacity

Up to 576 10GbE ports

Up to 864 1G ports


High density 10GbE w/future 40G/100G

HP 12500 highlights

Mid-plane designed for investment protection using CLOS Architecture

• Today: up to 720 Gbps FD • Ready for 40G/100G, FCoE

Resilient Virtual Switching Fabric (IRF) • Allowing multiple chassis (4) to operate as one device

Future availability of OSM Application Modules • Firewall, IPSec, Server Load Balancing, Network Management

Supports fully distributed L2/L3 Switching

Virtual output queues, 12,000 HW queues per module

256 MB I/O buffers for 10GbE ports

Ultra High Availability • Redundant /hot swappable fabrics, management /power supplies

• Hot swappable power supplies and fan assemblies

• Hot patch SW Upgrades, IRF based Hitless Upgrade

Low power consumption, max 9,780W, 7,237W typical • 50% less than direct competitor, Validated by Miercom

Full suite of Layer 2 and Layer 3 standard-based protocols

• 802.1D (RSTP/MSTP), 802.3ad (LACP), LLDP

• Supports IPv4 (RIP, OSPF, BGP-4 and IS-IS)

• Supports IPv6 (RIPng, OSPFv3, BGP+, IS-IS6, dual stack, tunneling)

Supports MPLS (LDP, L3 VPN, L2 VPN, MPLS, RSVP-TE)

Supports Graceful Restart (OSPF and BGP) based on RFC3623

Supports Resilient Ring Protection Protocol (RRPP) • 50 msec failure recovery over standard Ethernet

based rings

Flexible deployment options with choice of chassis • 4-slot (mid-size DC, aggregation), 8-slot chassis, and

18-slot chassis)

• Choice of AC or DC power


HP 12518 AC Switch chassis view

442mm

38 R

U

(168

6mm

)

9x Switching Fabrics

Fan Frame

Power Entry Module

Fan Frame

Power Module

Service Boards 6x modules

16x ports 10GbE each Non Blocking solution

Redundant Main Control


Today's network design challenge

• STP stops loops, but blocks paths - reducing effective bandwidth

• Configuration Intensive (switch by switch)

• Operational complexity for configuration change and support

• Slow convergence during failure can often impact business applications and users

Access

Core/Distribution

Active/Passive

Configuration effort

Operational Complexity

Active link

Blocked/standby link

Convergence time

User user/app experience

Legend


Configuration effort

Operational Complexity

Active link

Blocked/standby link

Convergence time

User user/app experience

IRF Platform Virtualization

Legend

How IRF meets this challenge

Access

Core/Distribution

Active/Active

Scalable Server Bandwidth All lin k s are act ive.

• Simplified/virtualized L2/L3 design and operations with a single IP and configuration file. No need for STP neither for VRRP. No MSTP design and operations headache. Switching and routing functions get virtualized as well as service modules.

• Allows multiple active paths for scalable performance

• Near instantaneous network recovery ensures business continuity and superior user experience

Logical View In Network

Management Tool

IRF physical topology to logical topology


HP IRF benefits summary HP IRF based designs STP based designs Competitive Solutions

Rapid fail-over Yes

Sub 50 millisecond - publically tested No

often measured in seconds

Varies by protocol

Design simplification – common across data center/campus core/edge

Yes Virtualizes up to 9 switches at the

edge, 4 in the core (12500) - common across layers/devices

No switch by switch configuration

No different protocols used at each layer and device type

Support for layer 2/3, MPLS, IPv6 protocols Yes L2 only Varies by protocol

Performance Very High Low High design/device specific

Geographic resiliency Yes up to 70 KM No L2 only - limited to site

Unclear varies by protocol

Overall administration effort and cost

Low No extra cost

High No extra cost

High often requires additional

license and hardware

Maturity > 10 years - recent


Data Center simplification with IRF Move from 3-Tier to 2-Tier network architecture

VMotion

Higher ports density and IRF allow to simplify DC network design, providing much better network performance for server virtualization (Vmotion)


IMC Data Center Core 1 Data Center Core 2

Networking/Security Servers Storage & Management

VCenter

FC Storage

Fibre Channel Switch 1

iSCSI Storage

Fibre Channel Switch 2

10 GbE

10 GbE

10 GbE

HP FlexFabric Reference Architecture


Rack servers Blade servers

W E Up to 75% of traffic

Up to 50% of ports interconnect switches; not servers or storage

Unnecessary layers add hops and latency

Slows performance of federated apps & VM mobility

N

S

Multi-tier Legacy Architecture Federated Applications and Virtualization changing traffic flows


Advanced architecture that improves server-to-server performance How HP FlexFabric is changing the rules

Access Switches with IRF, Virtual Connect

Higher performing, scale and switch/network reliability

Requires fewer layers and platforms—simplified lower cost design

IRF enables a single low latency hop between servers (for L2 and L3)

IRF = Intelligent Resilient Framework technology

Core Switches with IRF

Rack servers Blade servers

W E

Virtual Connect

IRF

IRF

Up to 75% of traffic

HP BladeSystem with Virtual Connect


Fibre Channel A

Fibre Channel B

C7000 Blade chassis With Virtual

Connect

A12518 Core A12518 Core

10 Gigabit Fibre

2-Tier Blade Server and Core- Rack View


HP C7000 Blade Series - Engineered to maximize Wire-once, change ready HP Virtual Connect Reduce server edge infrastructure up to 75%

Manage infrastructure in ½ the time HP Insight Management Reduce unplanned downtime by 77%

Power aware, energy optimized HP Thermal Logic 2X capacity without adding power infrastructure


Add I/O connections on-the-fly for virtual machines HP Virtual Connect Flex-10

HP ProLiant Server Blade

nics 1 2 3 4 5 6 7 8

LOM

66% less networking cost 75% less networking hardware 94% fewer cables 50% less power Over 3x the network bandwidth

4 redundant networks

8 ~ 0 + 2 nics require mezz and modules


HP Virtual Connect Flex-10 Overview


24

No license required to get full L2/L3 and IPv4/IPv6 dual stack features to ensure ease of use & out of box simplicity

HP’s licensing simplicity

NO License SKUs


Endura Servers

Full Multicast Implementation • PIM-Sparse-Mode • Multicast Source Discovery Protocol (MSDP) • IGMP v1,2 and 3

Pelco Video Surveillance over HP network Endura System Overview


HSL Distribution / Aggregation Layer Solutions


High Density 10GbE Ports & 40G Uplinks for Scalable DC and Aggregation

HP 5900AF-48XG-4QSFP

• Top of rack 1-RU switch for DC convergence

• Ethernet ports

– 48*10GbE SFP+ ports with 1GbE support

– 4*40Gb QSFP uplink ports (Each port can be split into 4 10GbE ports)

• High Performance

– 1.28 Tbps switching capacity

– 952.32 Million PPS throughput

– Integrated packet buffers (9MB)

• Hot swappable reversible DC airflow design, AC/DC hot swappable power options

• IPv6 performance in hardware and DCB protocol support

• IRF capable with 59xx devices (4 units)


IRF enables virtual chassis for the aggregation layer

HP 5900 - Port density & performance

• Simpler to design, to provision, to manage

• No STP, RSTP, MSTP, VRRP

• One IP and one configuration file

• 4 devices Stacking

• IRF with 10GE or 40GE

• Support cross-chassis link aggregation

• Models with 10GbE SFP+ or UTP technology High reliability

Simple management

High performance

Abundant functions

High scalability

Cost effective


HP 7500 Switch Series

7502 7503-S 7503 7506 7506-V 7510

Line Card Slots 2 2 3 6 6 10

Management/Fabric Slots 2 1* 2 2 2 2

Power Supply Bays 2 2 2 2 2 2

Max GbE ports 96 120 144 288 288 480

Max 10GbE ports 16 16 24 48 48 96

(*) 7503-S Fabric Modules include a Line Card


HP 7510 – Distribution Layer Switch Legacy Gigabit Uplinks

1. Dual SRPU load balance, 16RU horizontal chassis, 1+1 redundancy power, 10 service slots, hot swappable.

2. Switching capacity: 1152G*, packet forwarding: 780Mpps* (* with dual SRPU )

3. Automatic fan speed, low power consumption, low noise.

4. High GE and 10GE port density.

High GE and 10GE ports density


IRF: Intelligent Resilient Framework

• Up to 4 chassis per IRF-fabric

• Up to 8 10GbE or 40GbE physical ports per IRF-port

• Chain or Ring Topology

• All software features supported on an IRF-fabric

• Enhanced Mode for MPLS L3VPN, MPLS L2 VPN and VPLS

• BFC and LACP MAD


Migrating to Open Standards


• Enhanced Interior Gateway Routing Protocol (EIGRP) • Hot Standby Routing Protocol (HSRP) • Cisco Discovery Protocol (CDP) • Per-VLAN Spanning Tree Plus (PVST+) • Port Aggregation Protocol (PAgP)

Major Cisco Proprietary Protocols under use


Transition Strategies (EIGRP to OSPF) Flash Transition / Overlay 1. Configure OSPF on the network with a low routing priority. 2. Verify the routing information taking into account that the routes may not

match if the EIGRP design used multiple summarization points. 3. Raise the OSPF routing priority (in all the routers at once)

The routing table should now contain only OSPF routes.

4. Remove EIGRP from the network. Coexistence (Redistribution using existing C6500) •Configure mutual redistribution at the border between OSPF and EIGRP regions of the network.

– Coexistence may be temporary while different areas are transitioned.


Transition Strategies (HSRP to VRRP)

Flash Transition / Overlay • HPN supports the standards-based Virtual Router Redundancy Protocol (VRRP), not Cisco’s

proprietary Hot Standby Router Protocol (HSRP).

• This does not mean that either protocol is inferior. However, interoperability dictates a plan that implements standard protocols.

• Although HSRP and VRRP share the same concepts, they are incompatible. In other words, there is no co-existence between them.

• VRRP is fully supported by HPN, Cisco, and other vendors, and should be implemented wherever possible.

• IRF on the core eliminates the need for VRRP


Transition Strategies: CDP to LLDP

Coexistence

• Reason: Using Cisco IP Telephony Solutions

• LLDP-enabled device cannot recognize CDP packets and doesn’t respond to requests from Cisco IP phones for the voice VLAN ID configured on the device. This causes a requesting Cisco IP phone to send voice traffic without any tag to the device, which disables the ability to differentiate VOIP traffic from lower priority traffic (such as best effort).

• You can configure CDP compatibility mode on HPN devices to allow LLDP to receive and recognize CDP packets coming from Cisco IP phones, and configure the voice VLAN automatically. VOIP traffic can then be confined in the configured voice VLAN and differentiated from other types of traffic.


PVST was developed by Cisco to overcome limitations in the standards-based Spanning Tree Protocol (802.1d) • PVST is a Cisco proprietary protocol and it uses the also proprietary ISL VLAN identification mechanism

• PVST+ is a variation of PVST and uses the IEEE VLAN ID

PVST exist because when implementing STP in a multi-VLAN environment it is convenient to have several spanning trees in order to create VLAN load balancing.

PVST/PVST+ Weaknesses:

• Max of 128 STP instances on Cisco switches. Each VLAN will use an instance on PVST+, thus max of 128 VLAN per bridge region.

• Not suitable for DC ; Not enough VLANs; Huge impact due to flooding during frequent adds/deletes/updates.

• Uses more processing power on switch, more bandwidth and complexity

Per-VLAN Spanning Tree - PVST and PVST+


• HP Legal approved support of Cisco PVST+ interoperability mode in Comware (December 2010)

• PVST+ is being added as part of the HP software update of existing switches, and is in the base software of new switches.

• With both Core and Distribution layer operating on IRF mode, the access layer may use link aggregation control protocol (LACP) instead of PVST+ for traffic load balancing.

Transition Strategies (Migration from PVST+)


LACP is a open standard developed by IEEE, and it offers all the features of Cisco’s proprietary PAgP protocol. LACP is fully supported by both Cisco and HPN, and should be implemented where ever possible http://www.cisco.com/en/US/technologies/tk652/tk701/technologies_white_paper0900aecd80422c70_ns165_Networking_Solutions_White_Paper.html

Link Aggregation Control Protocol (LACP), and Transition Strategies (PAgP to LACP)

http://www.cisco.com/en/US/technologies/tk652/tk701/technologies_white_paper0900aecd80422c70_ns165_Networking_Solutions_White_Paper.html�

http://www.cisco.com/en/US/technologies/tk652/tk701/technologies_white_paper0900aecd80422c70_ns165_Networking_Solutions_White_Paper.html�


Storage and Backup Solutions


HP EVA Storage Application consolidation for traditional environments


EVA eliminates IT sprawl and stranded capacity for traditional applications Solution Virtual pools of resources for consolidation

Single point of management and control

Provide resiliency and flexibility

HP EVA Storage


EVA: automatically optimizes performance & capacity

Efficient Simple Adaptable

Easy to use Familiar Intuitive

Thin Virtual Green

Matrix Comprehensive HA

Data migration


HP EVA is simple

Familiar technology Still easiest to use Intuitive

20-30 Percent easier and more cost effective to use compared

to EMC and NetApp

One management platform for all your EVA storage arrays

Transparent integration of new features like Thin Provisioning


HP EVA is efficient

Optimized

Tiering balances $/GB and $/IOP

Thin

Reduce capacity requirements by 30%

Green

Reduce power and cooling costs at least 50%

Reduce the acquisition and operational cost of data center storage

Tier 0 (SSD)

Tier 1 (FC)

Tier 2 (Nearline) 50%


HP EVA is adaptable

High availability Matrix Data migration

Storage for ready-made converged infrastructure

Business continuity and disaster recovery solutions for your data

Lifecycle data management made simple with EVA P6000

and services from HP


Manage storage and VM resources in unison

EVA optimized for virtualization

VAAI offload improves VM performance

Command view

Insight control storage module for vCenter

VMware site recovery manager plug-in for P6000 EVA

Manage storage and VM resources in unison

Monitor the storage associated with VMware VMs

Enables VMware to build, manage and execute reliable

disaster recovery

Storage administrator

Virtual server administrator


HP storage reporting and automation within Matrix Operating

Environment service catalog

EVA integration for CloudSystem matrix and HP CloudSystem for private clouds

Traditional storage for cloud applications

Only HP has unified infrastructure with server and storage in a single application-centric service portal EVA P6000 improves CloudSystem Matrix storage utilization with its virtualized architecture and new thin provisioning


HP StoreEver MSL Tape Libraries


Unattended backup, DR and archiving for your business

HP MSL tape libraries

Meeting data growth challenges head-on

• Automating business data protection

• Scale up to 600 TBs of storage with LTO-6

• Even easier to use with ground-breaking LTFS

• Cost-effective consolidation – LTO-6 reduces the amount of LTO-4 media by 75%

• Keep pace with data growth with HP MSL Library Extender Kit

• Proactive drive, library and media monitoring with HP TapeAssure

• File management on tape for ease of use – HP StoreOpen with LTFS

• Integrated web-based remote management capability

• Added data security - AES 256-bit tape-based data encryption

• Encryption key management enabled through Data Protector Express software and HP Encryption Kit hp.com link to enterprise libraries

http://h10010.www1.hp.com/wwpc/us/en/sm/WF04a/12169-304612-304631-304631-304631.html�


Mid-Range Entry-Level Enterprise 1-20 slots 21-100 slots >100 slots

Now with LTO-6 for even higher scalability

HP StoreEver tape automation family

MSL Series

• Simple and low cost • Ideal for small servers or

small networks • DAS and LAN backup

• Manageable, flexible, reliable • Suitable for medium and large

businesses • DAS, LAN, entry-level SAN

backup

Autoloaders

EML E-Series

• Scalable, manageable, reliable

• Medium to large data centers • Medium to large SANs

ESL G3 series

• Higher capacity and performance points

• Scalable, dual robotics capable • Licensed features: automated

media pool (AMP), control path, data path failover etc.

• Enterprise data centers • Large-scale SANs


HP MSL Library Extender Kit is a mechanical mechanism that will connect two MSL libraries together allowing sharing of cartridges in new single, logical library.

Scalability: HP MSL Library Extender Kit

• Investment Protection − Leverage existing library investments to increase capacity and

performance as needed − Buy only the capacity you need today and then add additional

capacity or performance as needs increase − No need to relearn library manageability. Simply grow your

existing solution to meet new capacity and performance requirements

• Flexibility − Choose from a variety of MSL expanded options − MSL2024, MSL4048, or MSL8096 may be used together to create

a configuration that best meets the business need www.hp.com/go/automation

http://www.hp.com/go/automation�


Solution Offering


Storage and Backup Solution

HP P6550 EVA Dual Controller FC Array, with: • 02x HP M6625 2.5-inch SAS Drive Enclosure

• 42x HP 600GB 6G SAS 10K 2.5in HDD

• 08x HP 200GB 6G SAS SFF SSD

• HP X3800 G2 Network Storage Gateway

• Support SLA: 3y 6hr CTR Proactive Care

• Installation and Startup Service

02x HP 8/40 Power Pack+ 24-ports SAN Switch

MSL4048 + MSL8096, with Library Extender • 06x LTO-5 Fibre Channel Drives

• 144x Drive Slots