EMC Solutions For Optimized Epic EMREnvironments

EMC SOLUTIONS FOR OPTIMIZED EPIC
EMR ENVIRONMENTS
Redefining your environment for application services,
consolidation, and agility
ABSTRACT
Healthcare organizations face mounting challenges to “do more with less” and reduce
costs while implementing new EMR technologies such as the Epic Care EMR. To
address these challenges, the EMC XtremIO® all-flash storage solution provides
healthcare delivery organizations with a consistently high-performance, highly efficient
infrastructure to achieve more value from their Epic environment with fewer
resources. This white paper presents test results with XtremIO in an Epic environment
that show how XtremIO data services shrink the complete Epic data footprint by more
than 70 percent. Also covered are test results that demonstrate how XtremIO
performance exceeds Epic service level requirements by a wide margin. Ultimately,
this paper validates XtremIO as a solution that can help healthcare providers
accelerate their strategic objectives of improving patient care and decreasing costs.
February 2015
EMC WHITE PAPER
To learn more about how EMC products, services, and solutions can help solve your business and IT challenges, contact your local
representative or authorized reseller, visit www.emc.com/xtremio, or explore and compare products in the EMC Store.
Copyright © 2015 EMC Corporation. All Rights Reserved.
EMC believes the information in this publication is accurate as of its publication date. The information is subject to change without
notice.
The information in this publication is provided “as is.” EMC Corporation makes no representations or warranties of any kind with
respect to the information in this publication, and specifically disclaims implied warranties of merchantability or fitness for a
particular purpose.
Use, copying, and distribution of any EMC software described in this publication requires an applicable software license.
For the most up-to-date listing of EMC product names, see EMC Corporation Trademarks on EMC.com.
VMware is a registered trademarks or trademarks of VMware, Inc. in the United States and/or other jurisdictions. All other
trademarks used herein are the property of their respective owners.
H13860
2
TABLE OF CONTENTS
EXECUTIVE SUMMARY.............................................................................. 4 INTRODUCTION ....................................................................................... 5 EVOLVING EPIC AND INDUSTRY REQUIREMENTS ........................................................... 5 REDEFINING APPLICATION SERVICES, CONSOLIDATION, AND AGILITY
FOR EPIC ................................................................................................. 5 XtremIO Meets Epic Challenges Head On ...................................................................... 5 In-line, Scale-Up and Scale-Out Architecture ................................................................. 5 a rich set of always-on data services ............................................................................ 6 A CONVERGED INFRASTRUCTURE FOR EPIC ............................................ 7 ARCHITECTURE FOR CACHÉ AND COGITO CLARITY .................................. 8 VALUE ACROSS THE EPIC LANDSCAPE ..................................................... 9 PROTECTING EPIC PATIENT DATA ................................................................................ 9 XtremIO delivers HIGH Performance and Efficiency for Caché ........................................ 10 XtremIO Exceeds Epic SLAs ...................................................................................... 11 RICH SET OF DATA SERVICES ................................................................................... 12 HIGH THROUGHPUT FOR COGITO CLARITY ............................................ 13 HANDLING LARGE BLOCK READS FOR SQL SERVER Workloads ...................................... 14 VIRTUAL DESKTOPS DEMANDING TOP PERFORMANCE .......................... 14 GROWING Virtual desktop environments..................................................................... 15 SERVICES SERVERS: STRONG USE CASE FOR DEDUPLICATION .............. 17 CLINICAL ARCHIVING FOR LEGACY SYSTEMS ........................................ 17 PROFESSIONAL SERVICES ACCELERATE TIME-TO-VALUE ...................... 17 CONCLUSION ......................................................................................... 18 3
EXECUTIVE SUMMARY
Healthcare organizations face enormous challenges to reduce costs while continuing to invest in critical EMR technologies. These
challenges are expected to intensify as the government rolls out new requirements for meaningful use, including additional security
and data protection regulations.
The Epic EMR with its Caché and Cogito databases present unique workloads that are increasingly difficult for traditional disk-based
storage to handle without over-provisioning. Epic environments also contain numerous copies of the databases to support a wide
range of ancillary processes. This requires an even greater investment in traditional storage capacity.
The XtremIO all-flash storage solution addresses these challenges by delivering outstanding performance and efficiency for Epic
environments. It offers high throughput and consistently low latency that easily handle Epic's demanding workloads.
In addition, XtremIO provides an array of data services, such as instant snapshots, always-on, inline deduplication, and compression
to minimize the proliferation of data across the Epic infrastructure. XtremIO also delivers inline data services such as thin
provisioning, flash data protection, and encryption. What this means for an Epic IT environment is a dramatically reduced storage
footprint with less complexity and easier to manage, and reduced capital and operational expense.
This white paper presents technical evidence to validate how XtremIO provides breakthrough performance in Epic environments.
EMC conducted laboratory tests of three key Epic workloads: Caché, Cogito Clarity, and VDI. For each workload, Epic-specified
requirements, test environment descriptions, and test results are described. A sizing example based on an actual customer
implementation is also included.
XtremIO advanced data services such as deduplication, data compression, thin provisioning and on-demand writeable snapshots
deliver transformative potential for Epic IT environments. Such services demonstrate the ability to shrink the Epic data footprint by
more than 70 percent when compared to traditional storage infrastructure.
With XtremIO, healthcare providers will be able to achieve higher performance with fewer resources in Epic environments, all without
sacrificing security or data protection. Quite simply, XtremIO holds the key to increasing the value of Epic for patient care while
lowering costs, as well as achieving and maintaining regulatory compliance.
4
INTRODUCTION
The Epic electronic medical record (EMR) solution is a powerful, integrated system that enables clinicians to diagnose, treat, and
manage patient care with great precision and efficiency. The Epic solution also allows healthcare delivery organizations to meet
incentives established by the Health Information Technology for Economic and Clinical Health (HITECH) Act for adopting meaningful
use. As an EMR, Epic is one of the most important and significant investments a healthcare organization can make today.
EVOLVING EPIC AND INDUSTRY REQUIREMENTS
Epic is already a broad-reaching solution that places considerable demands on a healthcare provider’s underlying IT infrastructure.
As the industry evolves, those demands only are going to intensify.
Epic comprises several interdependent workloads: Caché database, Cogito analytics and reporting, virtual desktop infrastructure
(VDI), and numerous virtual servers for ancillary services. Each workload has specialized characteristics and stringent service level
requirements set by Epic. Traditional infrastructures often struggle to meet these demands and can become resource-intensive and
costly to maintain.
Epic environments are rife with duplicate data due to the need to maintain multiple database copies used for analytics, reporting,
test and development, and other processes. To accommodate this extra data, healthcare IT must continually add storage, further
increasing costs. Creating these copies is also problematic. Cloning the Caché database onto traditional spinning disk can take hours.
This forces administrators to clone during off-peak hours to avoid compromising production performance.
To handle growth more quickly and efficiently, healthcare organizations are moving toward the software-defined data center. This
transition will enable them to expand their environments easily by moving select workloads in and out of a hybrid cloud as
performance and capacity requirements surge and diminish.
Over time, there will be increased demand for data sources that provide valuable insights from data within and external to Cogito
analytics. These data lakes, which will consolidate multiple disparate data sources of both structured and unstructured data, will
become the standard for clinical research and help enhance patient care and operational efficiency.
In addition, increasing volumes of patient information, medical data, and business intelligence will require data- and patient-centric
security to protect healthcare providers from cyber threats and ensure their compliance with HIPAA, HITECH, and the Affordable Care
Act.
REDEFINING APPLICATION SERVICES, CONSOLIDATION,
AND AGILITY FOR EPIC
XTREMIO MEETS EPIC CHALLENGES HEAD ON
XtremIO is equipped to meet current and future challenges facing Epic environments. XtremIO is an all-flash, scale-out storage array
that is fully supported on Linux and VMware ESX operating environments, as well as UNIX environments, including AIX and HP-UX.
The XtremIO architecture is designed to provide the Epic IT environment with greater performance, scalability, efficiency, and data
protection.
The EMC solution delivers consistently high performance for Epic workloads, far exceeding Epic service level requirements and
providing a superior user experience. Data throughput, latency, and capacity utilization remain predictable and consistent, even as
system demands increase.
IN-LINE, SCALE-UP AND SCALE-OUT ARCHITECTURE
In a recent EMC-sponsored survey, hospital IT executives reported they expect data from their EMR environments to grow by an
average of 48 percent in 2015.
To manage this explosive growth, Epic users would benefit from the XtremIO unique architecture and scale-out design that delivers
investment protection with linear scalability to more than one million IOPS. XtremIO uses X-Bricks, which are building blocks that
can be easily clustered when additional performance and capacity are required. Unlike traditional scale-up flash arrays, X-Bricks
allow XtremIO to use all resources evenly across the cluster and provide long-term investment protection.
In this architecture, capacity continues to grow while consistent performance and throughput are maintained as nodes are added to
the system (Figure 1). Additional workloads are easily supported as new nodes are added. Such additional workloads may use
XtremIO data services and consistent performance to scale out and meet increasing IOPS, capacity and bandwidth requirements.
5
Figure 1. XtremIO Scale-Out, Scale-Up Architecture
A RICH SET OF ALWAYS-ON DATA SERVICES
XtremIO offers a rich set of always-on data services. For example, inline compression and deduplication shrink data volumes
dramatically to reduce storage capacity for Epic. (Figure 2) Unlike other solutions, XtremIO with not write duplicate data to disk
regardless of intensity of demand. This eliminates the need for post-processing to achieve data reduction. In addition, XtremIO does
not perform any system-level garbage collection. Instead, XtremIO uses advanced SSD functionality to continuously handle “garbage
collection” and completely avoid system performance penalties.
Figure 2. XtremIO Data Deduplication and Compression
In addition, XtremIO incorporates advanced security data-at-rest encryption. This capability allows healthcare organizations to
comply with evolving HITECH requirements without investing in additional technology. XtremIO also features agile, writeable
snapshots for generating copies of the Caché database while maintaining high performance of production workloads. XtremIO creates
snapshots, avoiding metadata bloat and the need for extra storage space. This empowers healthcare organizations to create and use
copies of Epic data as needed without consuming additional storage capacity or IT resources for management.
6
While delivering these powerful capabilities, XtremIO also helps healthcare organizations reduce total cost of ownership. The system
requires fewer drives to do the same work as traditional spinning disk, decreasing physical storage footprint. XtremIO is simple to
configure and manage, without the need for tuning or extensive planning. As a result, IT spends less time and money maintaining
operations and achieves significant operational savings.
A CONVERGED INFRASTRUCTURE FOR EPIC
To optimize an Epic environment EMC recommends a converged infrastructure. Epic users rely on VCE Vblock systems from EMC for
the fastest deployment of infrastructure and applications, the highest application performance and availability, and the lowest TCO.
The Vblock 540 System is the industry’s first all-flash-based converged infrastructure, incorporating the next-generation Cisco
Unified Computing System and ACI-ready network, and industry-leading VMware virtualization. With Vblock systems Epic users can
we simplify integration into existing and emerging management frameworks, providing complete system context of configuration and
health to introduce a new era of converged operations for IT infrastructure. Through focused integration with leading applications a
Vblock converged infrastructure simplifies application deployment and operations to deliver data protection, virtualization, and
simplified management of pools of storage resources across the healthcare enterprise.
EMC’s architecture for Epic is depicted below in Figure 3.
Figure 3. EMC Enterprise-Wide Architecture for Epic
Vblock systems are engineered by VCE working in concert with Cisco, and VMware, are manufactured and delivered as an integrated
product, and are supported as one product by VCE.
Although EMC recommends Vblock Systems because they enable fast deployment of Epic infrastructure and applications, and
enhance application performance at a low total cost of ownership, Epic users always have the option of building and supporting their
Epic infrastructure with components of their choice.
7
ARCHITECTURE FOR CACHÉ AND COGITO CLARITY
EMC has developed an XtremIO architecture that meets all of Epic's specifications and best practices for performance, high
availability, and recoverability. Because data in the Epic Caché database is irreplaceable and cannot be regenerated, utmost data
availability must be assured.
Figure 4. XtremIO Architecture for Epic
To protect against catastrophic data loss, a secondary data recovery methodology must be in place. To support the greatest level of
data resiliency, Epic strongly recommends storage of the Caché database on separate media from the associated journal files to
ensure data protection and recoverability. The reference architecture shown above includes two XtremIO arrays in the primary data
center. One array is dedicated to the production Caché database, report shadow, support, clones, and virtual machines (VMs) for
services servers. The second array is a multipurpose cluster for the journal files, Cogito Clarity reporting database, Cogito Data
Warehouse, and various test environments. The XtremIO arrays could be part of a Vblock 540 configuration or stand-alone as
depicted in figure 4.
For business continuance, this architecture includes an EMC storage array in a secondary data center. To simplify managing the
environment, healthcare providers can use the EMC VPLEX® storage virtualization solution depicted in the architecture above to
present multiple XtremIO arrays as a single array while achieving the separation Epic recommends.
Healthcare IT teams may choose to use EMC ViPR® software-defined storage management solution to centrally automate
provisioning and delivery of XtremIO resources. To help increase operating efficiency and streamline management, EMC also
provides ViPR as a single management interface for all of the arrays in the Epic infrastructure. ViPR aggregates physical storage into
virtual arrays and provides administrative control to manage XtremIO and Isilon storage, compute and networking resources, as well
as Epic databases. ViPR also provides direct integration to VMware for complete management and monitoring of the VDI and virtual
machine environments.
8
VALUE ACROSS THE EPIC LANDSCAPE
PROTECTING EPIC PATIENT DATA
For local protection and high availability, Epic requires mirroring of the Caché database using built-in Epic replication technologies.
EMC augments Caché mirroring with EMC VPLEX virtual storage to achieve near active-active operations across campus data center
sites for numerous functions. VPLEX enables two separate campus-based data centers to act as one, providing simultaneous access
to storage from either site. If connectivity to Epic is lost at one data center, it automatically continues to run from the other data
center without disrupting clinical services.
VPLEX also provides the flexibility to relocate workloads from one site to another and minimize planned downtime for maintenance or
upgrades. Adding EMC RecoverPoint® to the VPLEX environment for non-Caché environments provides continuous data protection by
enabling providers to restore data to a specific point in time. (Figure 5) This powerful combination of VPLEX and RecoverPoint
provides healthcare organizations with a highly efficient means to recover instantly data for compliance purposes and strong
protection against data loss.
Figure 5. Continuous Protection of Epic Environments with VPLEX and RecoverPoint
In addition, EMC provides a complete enterprise backup solution with The EMC Data Protection Suite and Data Domain®
deduplication storage systems. (Figure 6) Enabling daily full backups and single-step recovery, Data Domain provides healthcare
organizations with a single solution for backing up all of their Epic data, as well as data residing in other environments.
Data Domain deduplicates data during the backup process and can reduce backup volumes 10-30 times, as well as save space and
lower costs. Data Domain Replicator software provides safe offsite storage that eliminates the need for tape. Data Domain is ideal for
large and growing Epic environments because it easily scales by simply adding appliances.
9
Figure 6. EMC Recommended Architecture for Backing Up Epic Production DB
The Epic application uses a wide variety of data sets. In addition to the production database, which is based on Caché, Epic also uses
SQL or Oracle data warehouse and reporting instances, Web Binary Large Object (BLOB) file space and various Citrix and VMware
data sets that require a combination of full and incremental backups.
Epic’s diverse data types and backup requirements require a comprehensive backup and recovery solution designed to meet the
most stringent SLAs. EMC’s Data Protection Suite software and EMC Data Domain provide a complete enterprise solution with a rich
feature set and flexibility to address each unique use case. The combination of multi-streaming and Data Domain Boost distributed
processing delivers the robust performance needed for both backup and recovery of the Epic environment.
XtremIO DELIVERS HIGH PERFORMANCE AND EFFICIENCY FOR CACHÉ
One of the primary use cases for XtremIO is the Caché database—the central source for patient data that feeds all other processes
and workloads in the Epic environment. The database runs on physical UNIX or is virtualized with Linux using VMware vSphere.
As a post-relational, transactional database, Caché presents several challenges, which are outlined in Table 1. First, the database
has nearly 100 percent random reads, creating highly unpredictable I/O on the storage array. In addition, Caché has a write burst
that is flushed out to disk every 80 seconds. To ensure an acceptable user experience, Epic requires this write burst to complete in
less than 30 seconds. Epic also specifies an average write latency of less than one millisecond while sustaining a read latency of less
than 15 milliseconds.
Read/Writes
Must assume 100% random reads/writes
Average Read Latency
Less than 15 milliseconds
Average Write Latency
Less than one millisecond
Write Bursts
Flushed every 80 second; must be completed in 30 seconds to ensure positive user experience
Destaging
All Flash Array must handle write flushes without destaging
Table 1 Epic Performance Requirements for Caché database
To meet end-user performance requirements with traditional spinning disks, healthcare IT often needs to over-provision the Epic
storage environment using traditional spinning media arrays. As the Caché database grows and mechanical disks are added to keep
up with growing demand, healthcare organizations often encounter performance and latency problems. This leads to an inconsistent
PHI availability and user experience, as well as increasing capital and operating expenses.
10
XtremIO EXCEEDS EPIC SLAs
The all-flash XtremIO array is ideal for handling highly randomized workloads such as Caché. For validation, EMC created a test
environment comprised of two XtremIO X-Bricks. EMC also configured the environment with four host bus adapters (HBAs) and two
front-end ports on the XtremIO array to support the Caché database I/O. EMC PowerPath software provided automated data path
management and load balancing for the test environment. Finally, EMC used Epic's GenIO utility to accurately simulate production
I/O of Caché workloads.
Running Caché workloads on XtremIO delivered consistently high performance and low latency levels that far exceeded Epic's
established SLAs. And with its thin provisioning capability, XtremIO achieved these results without over-provisioning.
In handling Caché's random reads, XtremIO thoroughly eliminated any latency issues. While the maximum read latency Epic will
tolerate is 15 milliseconds, XtremIO did not exceed 1.3 milliseconds during the tests. Figures 7 and 8 show EMC lab results testing
Epic GenIO running on XtremIO 3.0 for read and write latency ranging from zero to more than 70,000 IOPS.
Figure 7. XtremIO Results for Caché Read Latency
Figure 8. XtremIO Results for Caché Write Latency
XtremIO consistently maintained extremely low latency even as the tests scaled from zero to more than 70,000 IOPS. This provides
healthcare organizations with an enormous amount of headroom for growth. In addition, XtremIO is able to deliver these
performance results with active snapshots running. This means that Epic users can create clones during the busiest production hours
without compromising performance. They also can refresh their snapshots more frequently at any time of day or night.
Similarly, XtremIO performance also exceeded Epic's write SLAs for Caché. (Figure 8) EMC ran tests in the range of 3,000 to 40,000
IOPS per second with latency measured in microseconds.
11
To meet Epic's established SLA, write latency cannot exceed one millisecond. EMC's tests showed that XtremIO delivered consistent write
latency as IOPS scaled, never exceeding 200 microseconds or 0.2 milliseconds. This opens opportunities for healthcare organizations to
explore extended clusters or other infrastructure designs for enhanced data mobility and disaster protection. Healthcare providers can
reduce risk, downtime, and cost of their Epic infrastructures without concerns about exceeding the write latency SLA.
RICH SET OF DATA SERVICES
Not only does XtremIO provide exceptional performance, but it also offers value with always-on data services, such as deduplication and
compression. While Epic specifies a minimum of six full copies of the Caché database in the EMR environment, EMC has found that most
healthcare organizations use at least 12 full-sized copies. EMC has developed a set of rigorous sizing rules based on sizing methodologies,
lab testing, and field validation of nearly 200 Epic customers. The sizing comparisons in Tables 2, 3, and 4 is based on these inputs.
By employing inline deduplication and compression, XtremIO dramatically reduces the total amount of storage capacity and
associated cost required for all of these copies. For example, EMC compared sizing for a 9.3 terabyte Caché database between
traditional spinning disk and XtremIO. Traditional disk technology required 179 terabytes to support the original database, including
all of the copies and required Epic storage pools. (Table 2)
Disk Sizing
Epic Pool
Number
Pool 1
Pool 2
Pool 5
Pool 6
Pool 8
Epic
Capacity
Required
Epic Storage
Function
Disk
Model
Min.
Drive
Count
Array
Capacity
(GB)
Production database
Reporting environment
data, production journal
Clarity and production
release & testing
environments
9,600
300GB 15K
Production
164
22,009
13,600
300GB 15K
Production
112
15,030
59,300
600GB 15K
Production
164
66,026
VMs, BLOB, Archiving
Clones for nightly
backups
33,850
300GB 15K
Production
188
37,844
23,600
600GB 15K
Production
68
27,377
Production Array Totals - Prod on FC
178,741
696
Table 2: Epic Traditional Disk Sizing Table
In contrast, XtremIO deduplication supports that same 9.3 terabyte database and all of its copies with just 46 terabytes of all-flash
capacity for the same configuration as defined by Epic from an Epic Preliminary Hardware Configuration Guide. (Tables 2 and 3)
That translates to a 73 percent reduction in storage footprint. The degree of consolidation of storage capacity requirements between
traditional storage and XtremIO all-flash arrays varies by implementation and each end-user’s unique environment. Tables 3 and 4
are based on a recent XtremIO deployment for Epic.
Production Database Cluster
Epic Pool
Number
Pool 1
Pool 2
Pool 5a, 8
Pool 6a
Epic Storage
Function
Epic Capacity
Required (GB)
Consumed
Capacity (GB)
Production Database
11,500
7,667
Reporting Data
11,700
1,560
Clones, support and backup
33,200
4,427
6,380
2,552
Multipurpose VMs/Service Servers
Total Capacity Demand
Total Capacity Consumed
62,780
16,205
Table 3. Epic XtremIO Cluster Sizing with inline deduplication and compression
12
Multipurpose Cluster
Epic Pool
Number
Epic Storage
Function
Pool 2a
Production journal/application files
Pool 2b
Pool 5b
Epic Capacity
Required
Consumed
Capacity
1,900
1,140
Reporting data
11,700
7,800
Non-production Caché
21,000
1,750
Pool 5c
Cogito Clarity data warehouse data
13,600
9,067
Pool 5d
Cogito Clarity data warehouse test data
13,600
1,813
Pool 5e
Test/Train database
Pool 6b
Citrix VDI
Pool 6c
Web BLOB Server Data
Total Capacity Demand
1,500
800
17,600
3,520
3,600
103,700
3,600
Total Capacity Consumed
29,490
Table 4. Epic XtremIO Cluster Sizing with inline deduplication and compression
As additional testimony to the value of XtremIO, a customer using XtremIO for Epic discovered that two additional snapshots of the
production Caché database required virtually no additional space when created. The healthcare provider had allocated 16 terabytes of
space for the seven-terabyte production Caché database for growth. With thin provisioning and the addition of these two snap copies, a
total of only 3.4 terabytes of storage capacity was required--a space savings of 80 percent. The customer also expects the overall
efficiency rate of 19:1 to increase as more snapshots are generated. These powerful results come from XtremIO application services,
such as thin provisioning, in-line deduplication and compression.
HIGH THROUGHPUT FOR COGITO CLARITY
Another prime use case for XtremIO is Cogito Clarity, Epic's integrated analytics and reporting environment. Cogito Clarity is a
relational database, typically using Microsoft SQL Server. Historically, Cogito Clarity runs on physical servers, but more recently
healthcare providers moving toward virtualized environments for this database environment.
The database supports the Cogito Clarity reporting engine, which generates thousands of Epic Care reports from general ledger to
patient census and drug administration. It also feeds the Cogito Data Warehouse, which combines Epic data with data from other
systems in the enterprise.
The Cogito Clarity environment has two key workloads that are different from Caché and present their own challenges to the
underlying infrastructure. The first workload is Extract, Transform, and Load (ETL), which collects data from a copy of the Caché
database, transforms it to match a SQL Server format, and writes data to the Cogito database.
Data writes in the ETL process typically run at 25 to 500 megabytes per second or greater depending on the size of the environment
and server capabilities. Consequently, throughput becomes one of the biggest challenges in traditional spinning disk storage systems
due to physical limitations of a mechanical device.
The second key workload is reporting. Typically, the heaviest impact on the Cogito Clarity database is batch reporting, which often
runs three to four hours, or even longer, performed nightly. During this process, database reads can push toward 800 megabytes or
even one gigabyte per second. This level of read intensity can quickly overwhelm traditional disk-based storage systems.
EMC tested these SQL workloads on XtremIO to determine how well the all-flash array could handle the read/write demands. The
test environment utilized an XtremIO 2X-Bricks cluster. EMC PowerPath software again provided automated data path management
and load balancing to further optimize the test environment.
13
HANDLING LARGE BLOCK READS FOR SQL SERVER WORKLOADS
With six X-Bricks, XtremIO ingests data at a rate of up to 4.8 gigabytes per second. This rate is five to six times what is typically
required and easily handles the largest Cogito Clarity workloads. The tests also showed that XtremIO excels in handling the large
block sequential reads that Cogito Clarity demands.
In general, XtremIO has proven that it is very effective for heavy SQL Server workloads. EMC conducted a series of tests running
nine SQL Server workloads concurrently on XtremIO. Even when workloads drove more than 4,200 transactions per second and
200,000 IOPS, XtremIO maintained high performance with consistent latency of less than one millisecond. (Figure 9)
Figure 9. XtremIO Maintains Low Latency Under Heavy SQL Server Workloads
VIRTUAL DESKTOPS DEMANDING TOP PERFORMANCE
Virtual desktop infrastructure (VDI) is becoming a critical environment for Epic as more clinicians and administrators seek greater
mobility. The biggest challenge for most VDI environments is controlling the cost per desktop while delivering high performance and
scalability.
Epic supports two VDI approaches: VMware Horizon View or Citrix XenApp. Horizon View allows organizations to deliver virtual
desktops running the Epic Hyperspace application directly to the local user device as a single solution. XenApp provides remote
delivery of Hyperspace but also needs to run on a VMware ESXi host to support the scale necessary required by many Epic
environments. Another option is a combination of XenApp and Horizon View to present the Hyperspace application and further
increase the agility and mobility of a virtual desktop environment.
XtremIO provides the I/O performance and scalability required to deliver a smooth and consistent end-user desktop experience for
both Horizon View and XenApp environments. The I/O requirements for XenApp are higher than Horizon View and range from 20 to
40 IOPS per desktop, a level of performance that is difficult to achieve with traditional spinning disk. With its scale-out architecture
and deduplication capability, XtremIO can sustain high performance and low latency for up to 3,500 Horizon View desktops or as
many as 2,500 XenApp servers per X-Brick. The appropriate number of supported desktops for any deployment will vary depending
on the configuration.
14
GROWING VIRTUAL DESKTOP ENVIRONMENTS
EMC tested XtremIO with Horizon View workloads in a laboratory setting on a VCE Vblock infrastructure configured with two XtremIO
X-Bricks. (Table 5)
VM Attribute
Configuration
Desktop Operating System
Microsoft Windows 7 (Ultimate), 64-bit
VMware Virtual Hardware
Version 8
VMware Tools Version
9.4.10.2068191 (Up to Date)
Virtual CPU
2vCPU
Virtual Memory
2GB (Reserved)
OS Page File
2048MB
vNICs
1
Virtual Network Adapter Type
VMXNET3
Virtual SCSI Controller
LSI Logic SAS
Virtual Disk (VMDK)
32GB
Virtual Floppy Drive
In Place
Virtual CD/DVD Drive
In Place
Optimizations
VMware Windows7 NoPersona
Installed Software
Microsoft Office 2010
Horizon View Agent v6.0
LoginVSI Target SW v4.1.2
Adobe Acrobat Reader 10, Doro PDF Printer, FreeMind, Notepad
Table 5. XtremIO Test Configuration for Horizon View
The tests demonstrated outstanding desktop performance on XtremIO with strong deduplication and compression results. On
average, storage latency for 100 virtual desktops was consistently under one millisecond. Desktop pools were deployed and all 1,000
desktops were powered on and ready to use in just 55 minutes. (Figure 10)
Figure 10. XtremIO Delivers High Performance to 1,000 Virtual Desktops
15
In addition, XtremIO achieved deduplication and compression rates expressed in ratios described in Table 6.
Multiple test runs
produced a range of deduplication and compression metrics. This table shows these rations with the range of test results.
Clone Type
Deduplication
Compression
Full
5:1 to 15:1
1.5:1 to 2.2:1
Linked
2.5:1 to 4:1
1.4:1 to 2.2:1
Table 6. Guidelines on XtremIO Inline Data reduction Ratios for VDI
XtremIO also proved to be up to five times more efficient than traditional storage in recomposing virtual desktop pools. In a 1,000
desktop recompose in the XtremIO lab, EMC recorded approximately 30,000 IOPS at 850 megabytes per second, as depicted in the
screenshot below. The 30,000 IOPS are mixed across multiple I/O sizes. (Figure 11)
Figure 11. Typical Performance Profile for 1,000 Medium (Knowledge Worker) Users
In this lab test, EMC observed an average of 850MB/s with peaks in excess of one gigabyte per second. (Figure 11). The 850
megabytes per second performance level equates to more than 100,000 8kB I/Os on the array backend, but the XtremIO reporting
collates all I/O requests regardless of transfer size into a single front-end IOPS figure. This test was conducted with a single data
store as a performance experiment.
Figure 12. Typical Performance (Latency) Profile for 1,000 Medium (Knowledge Worker) Users
During a 1,000-seat recompose, XtremIO supported 70,000 IOPS at 1,000 megabytes per second. Desktop pool maintenance
operations also had little impact on existing user populations. This would eliminate the need for healthcare IT staff to schedule these
16
activities on nights and weekends. By delivering virtual desktops so efficiently, XtremIO helps reduce the cost per desktop
significantly.
SERVICES SERVERS: STRONG USE CASE FOR DEDUPLICATION
Epic Services Servers run a variety of Epic functions, from basic print and Web services to Federal Information Processing Standard
(FIPS) reporting on controlled substance distribution. A typical Epic environment may contain anywhere from 20 to 50 services
servers. Each services server is a virtual machine running VMware combined with Microsoft Windows. Data is comprised primarily of
the Microsoft Windows operating system and tends to be highly repetitive from one services server to another. That makes the
services server environment an ideal use case for XtremIO strong deduplication capabilities.
In addition, services servers have a demanding I/O profile similar to the virtual desktop infrastructure built on XenApp and VMware.
With its clean-sheet, scale-out design that maximizes benefits from flash, XtremIO sustains high throughput with greater consistency
and efficiency than traditional storage infrastructures.
In testing, EMC has determined that XtremIO will support I/O requirements of 2,500 virtual services servers on a single X-Brick. EMC
also has documented that XtremIO provides deduplication rates of approximately 65 percent, substantially reducing the amount of
storage space required for VM environments.
CLINICAL ARCHIVING FOR LEGACY SYSTEMS
Today’s Epic EMR solutions and the older legacy systems that share the same IT environment are generating increasing amounts of
all types of patient-related information, such as clinical and administrative documents, voice recordings, and medical images, and
more. Managing this data growth is a challenge for IT.
In addition, as organizations transition to Epic, the older clinical systems will become obsolete and often need to be maintained in
read-only mode to continue providing clinicians access to their data. This makes it more difficult for healthcare providers to
decommission these legacy systems and realize the cost savings, reduced complexity and decreased risk that would typically follow
after removing them from the environment.
EMC takes a holistic approach to supporting an enterprise Epic environment, knowing that any “weak link” in the free flow of patient
information to authorized caregivers can degrade the quality of care delivery. The EMC Documentum Clinical Archiving Solution is
built specifically for Epic EMR environments and addresses the information management and cost challenges of managing and
maintaining legacy systems. This Web-based application launched from Epic Hyperspace enables users to find and view archived
records, both structured and unstructured data, including clinical documents and medical images. Clinical Archiving also gives Epic
users on-demand access to all archived patient records, documents and images—eliminating the need to navigate multiple systems
and user interfaces. Clinicians gain immediate access to Hyperspace for a complete view of patient history, diagnosis and treatment.
Based on XML standards, Clinical Archiving provides healthcare providers with scalable and seamless interaction with their Epic EMR
solution and other IT environments by using native Epic APIs and industry standards. Documentum Clinical Archiving runs on the
Documentum Content Server, which is supported by a variety of EMC storage platforms.
PROFESSIONAL SERVICES ACCELERATE TIME-TO-VALUE
EMC Professional Services staffs a team of Epic and IT experts who have delivered 200 successful IT environments for Epic to
healthcare providers. The EMC Services Delivery team ensures that deployment of EMC infrastructure solutions abide by the most
up-to-date best practices and Epic-specific configuration rules.
Benefitting from extensive EMC eLab testing and collaboration between EMC and Epic engineering, EMC Professional Services have
the skills, knowledge and resources to plan and execute successful Epic deployments. Compared to Epic’s other infrastructure
vendors, the EMC Professional Services Industry Verticals Team provides an unparalleled level of service for all Epic deployments.
This team helps to support the solution sizing and creation of policies, procedures and tools used in delivery and deployment.
As part of its service offerings, EMC provides Epic-specific configuration, deployment and go-live activities for customers running Epic
in XtremIO environments. In addition, EMC Professional Services have established SKUs and bundles to streamline infrastructure
deployments supporting the Epic application.
17
CONCLUSION
XtremIO delivers game changing advantages to Epic EMR environments. It directly addresses the most pressing challenges
healthcare providers face: lowering costs and delivering new application services without performance impact, all while improving
patient care and enabling compliance.
Where traditional storage systems with mechanical disk drives may struggle to handle Epic's unique workloads, XtremIO excels. In
fact, XtremIO far exceeds Epic's SLAs, delivering consistently high performance even under the most demanding conditions. That
means clinicians can count on a reliable, highly-responsive EMR to support fast, efficient patient care.
Outstanding XtremIO performance, scalability, and efficiency also help healthcare organizations drive greater use of analytics to
better manage population health, optimize clinical quality, and mitigate risk. The all-flash array also provides ample headroom to
take on an increasingly mobile healthcare workforce.
Beyond raw performance and efficiency, XtremIO offers a wealth of data services that further enhance Epic. XtremIO dramatically
reduces the storage capacity required for Epic through always-on inline deduplication and compression. This is a prime advantage by
reducing capital expenses. The built-in data at-rest encryption capabilities of XtremIO enable healthcare organizations to comply with
new HITECH security regulations without additional investment.
In addition, instant snapshots created with XtremIO eliminate the manual, time-consuming process of creating database copies.
Instead, healthcare organizations can create snapshots any time of day or night without impacting production EMR services or using
valuable storage space.
EMC ViPR integrated in VMware Operations simplifies management of the entire Epic infrastructure with a single, centralized interface
that spans storage, compute, virtual machines, and networking resources. This not only reduces operating costs, it also frees
healthcare IT resources for strategic projects that address emerging HIPAA requirements and the HITECH Act. Ultimately, this saves
even more money by avoiding costly penalties for non-compliance.
From the data center to the bedside, from deployment and delivery to legacy application decommissioning and archiving, XtremIO
and the entire EMC Federation portfolio deliver performance, efficiency, scalability, and availability to help healthcare organizations
achieve the most value from their Epic investments.
18
Download PDF
Similar pages