This presentation discusses continuous availability and non-disruptive data mobility with EMC VPLEX.
In summary, there are a lot of data points. A few key points we’ve called out are:
64% consider data protection to be critical to their organization’s success, still 64% suffered downtime or data loss in last 12 months
More data protection vendors means more data loss and more spending on data protection
87% of businesses are behind the curve for data protection maturity and 71% of businesses are not fully confident in their ability to restore apps/data
The majority of businesses have suffered data loss in the last 12 months.
But, those businesses that are ‘ahead of the curve’ and have the most mature data protection strategies are the least likely to suffer incidents.
We also asked respondents to quantify the cost to their business of downtime.
The figures for downtime supported the same story.
This slide shows how that is played out technologically.
Just 9% have built their strategy on an active-active strategy, and the results show that they have fewer disruptions. 15% of these organizations experienced data loss with an active-active strategy.
In comparison…
Over 40% of businesses still rely on backup for their data protection. 31% of these organizations experienced data with a traditional backup strategy only. This means that just relying on backup means companies are 2x more likely to lose data than have a continuum of data protection.
EMC thinks of this as a data protection continuum that spans protection tiers to provide everything from continuous availability to replication to backup and archive. These tiers all need to work together in a complimentary fashion which is why EMC has brought together our availability, replication, backup and archive solutions under the Data Protection and Availability division. Today, we will be discussing the left-side of the spectrum, where RPO/RTO is zero for continuous availability and RPO/RTO is minutes/seconds for local and remote replication.
When considering Disaster Recovery planning for mission critical applications there are two key metrics that are typically used to determine recovery:
Recovery Point Objective (RPO); the measurement of data loss in time from before the disaster event when a recovery is executed. Effectively, RPO is the amount of data you are willing to lose in the event of a disaster. The more critical the data, the lower the RPO should be.
Recovery Time Objective: (RTO); The measurement of time it takes after the data has been recovered that it takes to restart the applications so that customers can resume production operations. Effectively, this is amount of production downtime that you can sustain after an event has occurred. The more critical the application, the lower the RTO should be.
As more critical applications are virtualized and as data centers move towards Public, Private and Hybrid Cloud environments, expectation for both RPO and RTO will be compressed. It is important for you to know that not all applications have the same recovery requirements and that a “one size fits all” approach to Disaster Recovery and Business Continuity no longer applies.
Putting all of that success together, we continue to bring in new customers at a fantastic pace.
VPLEX:
In 72% of the top 200 EMC enterprise accounts
Over 2000 customers
Over 5200 units deployed
Over 5-9s uptime
Over 68M run hours
Over 500 PB virtualized
RecoverPoint:
In 85% of the top 200 EMC enterprise accounts
Over 11,00 customers
Over 21,000 units deployed
Over 320M run hours
Over 1 EB protected
It’s easy to get started. No barriers to entry
New strategies for Business Continuity include establishing Active-Active environments, either within the data center or across data centers.
EMC’s VPLEX storage virtualization platform allows you to create an Active-Active environment and deliver Continuous Availability.
Here’s an example of how VPLEX works:
We have two cluster server environments. Maybe one VMware HA cluster and the other Oracle RAC or IBM’s Purescale.
Each of these clusters connect to a single storage array, is a single location.
<Click> We introduce VPLEX into the environment and each of the volume are virtualized by VPLEX.
<Click> From this, VPLEX creates what are known as Distributed Virtual Volumes that go across arrays and store an exact copy of the volume in the other array. Each copy of data is fully read/write accessible by the Cluster Server and creates what is known as an Active-Active environment.
EMC VPLEX enables Active/Active distributed DCs.
VPLEX is a storage virtualization solution. There are many virtualization solutions in the market place today, however the key differentiator is that VPLEX was designed from day one to enable multi-site storage virtualization.
VPLEX is a hardware and software solution that can be used to create an abstraction layer between hosts and storage arrays. From a storage perspective VPLEX looks like a host and you can present Physical LUNs from the storage array to VPLEX. From a host perspective VPLEX looks like a storage array and you present virtual LUNs from VPLEX to the host. By breaking the link between the host and the physical storage array it enables storage virtualization use cases like non disruptive data mobility across array.
In the context of the conversation, it is possible to take another datacenter, with another physical storage array and a completely separate, resilient VPLEX cluster. The VPLEX Clusters can then be federated together. Federation is the process of creating a communication path between the VPLEX clusters. This can be done over FC or IP, however the RTT between the DCs must be lest than 5ms and there must be enough bandwidth to sustain synchronous replication. You do not need to merge fabrics between sites.
Once federated it is possible to create a distributed virtual volume. This of the distributed virtual volume a bit like a RAID 1 mirror, reads and writes can occur at both sites and VPLEX consistently mirrors the data across storage arrays. One advantage of this architecture is that it enables 100% storage availability. In the event of a problem with an array, there is not requirement to fail application from site A to site. Read and writes continue to be served from either VPLEX cache or the remote array.
The key consideration in regard to distributed virtual volumes is that they have exactly the same storage and LUN identifiers across both sites and are concurrently RW accessible. Why is that important..
With VPLEX the solution is simplified and additional availability is enabled.
With VPLEX Oracle RAC can be configured across sites with none of the normal complexity and the storage I/O from both nodes can be leveraged for a single set of LUNs.
Existing VPLEX and RecoverPoint products, offering continuous availability and operational and disaster recovery.
VPLEX
Active-Active
Transactions must be running and stable at two locations within Metro distances, mostly financial institutions
Provide Continuous Availability With VPLEX Array Mirroring
Mirror data from XtremIO to XtremIO to protect against the loss of an array or even an entire site.
Move Workloads To and From XtremIO Without Disruption
Application data can be stored on spinning disk and non-disruptively moved to XtremIO when performance requirements demand it
Expand X-bricks Without Disruption With VPLEX
Putting VPLEX between hosts and storage allows IT to add XtremIO capacity and performance by adding X-bricks without disrupting application data access
RecoverPoint
Remote replication with recovery to any point-in-time
Universal replication – any application, any distance, any array using VPLEX, any network
Supports bi-directional asynchronous replication over a Fibre Channel or IP network over any distance
Saves 90% of network bandwidth consumption thru Snap and Replicate capabilities, deduplication and compression
Ensures application-consistency
Enables application-level consistency and seamless re-starts between multiple arrays and multiple inter-dependent applications
Multiple configurations/sites
Fan-out capabilities for data distribution and dev/test environments
MetroPoint topology – industry-unique continuous availability with continuous disaster recovery with the ability to withstand a 2 site failure.
Field Proven
In 85% of the top 200 EMC enterprise accounts
Over 11,000 customers
Over 21,000 units deployed
Over 320M run hours
Over 1 EB protected
For customers requiring even further levels of availability, VPLEX Metro Express Edition can be used with RecoverPoint CRR to enable a third passive DR site in another out of region location.
There are two additional Use cases that without VPLEX would be hard to image.
Here’s an illustration of the time saved with VPLEX
EMC introduced VPLEX Virtual Edition in April, which allows deployment of VPLEX as a low-cost virtual appliance, bringing VPLEX continuous availability and data mobility to VMware-centric mid-tier organizations.
Screenshot of vSphere Web Client with VPLEX/VE plug-in installed.
Screenshot of the vSphere Web Client showing the VPLEX/VE metro site configurations, with 4 vDirectors per site.
Screenshot of vSphere Web Client, highlighting the VPLEX/VE storage administrative tasks exposed through vCenter. Tasks include:
Creating a New Distributed Datastore
Expanding an existing Distributed Datastore
Deleting a Distributed Datastore
Changing the site bias – or primary and non-primary site
Migrating Back-end Storage
Who is the for?
If you have 2 sites, VMware, iSCSI, VNXe
And you have Mission critical virtualized applications
VE Changes the game for you
Along with the VPLEX appliance, new option of VE – integrated with VMware, vApp alignment
And with VE you have the same great use cases for VMware that exist with VPLEX appliance
CA
Instant vMotion between sites
Instant LB between sites
Here are some examples we have discussed with potential customers:
Branch office workloads – Imagine a large metropolitan area where you may have multiple branch offices. You may have just a single standard rack of equipment at each branch, and with VPLEX/VE you can have all your IO synchronously written at the branch and a central data storage location. You’d have full read/write access from the central location and the branch office at the same time. You business systems could move as necessary.
Cruise ship HA – How about a modern mega-cruise ship? We’ve had customers looking at running separate server rooms at either end of the boat and providing HA between the two “data centers”.
Users will be able to download a trial version of VPLEX Virtual Edition software at no charge and no time limit for non-production environments starting on January 20, 2015. Installation and administration and management support will be provided through our active VPLEX Technical Community. Product documentation sets, Quick Start Guides, Installation Guides and How-to videos will all be provided for our users. And, they have the ability to collaborate with EMC experts and other developers.
Overcoming the management concerns with ViPR Controller and enabling a Data Protection Services Catalog
ViPR Controller enables you to manage heterogeneous storage environments, including EMC, Non-EMC and commodity storage. In fact, any block storage supporting the OpenStack Cinder Driver can be managed by ViPR controller. But what makes ViPR controller unique, is that unlike previous implementations of storage virtualization, ViPR Controller decouples the control path from the data path.
By abstracting the control path, storage management operates at the virtual layer, which gives customers the ability to partition a storage pool into virtual storage arrays.
This is analogous to partitioning a server into a number of virtual machines. Control path data services provide multi-tenancy, service cataloging, metering and monitoring across all arrays.
ViPR Controller does not sit in the data path for file and block stores. This ensures applications can access storage and all its underlying value and data services embedded in the storage arrays—without performance degradation. It also enables administrators to centralize data provisioning and data management tasks, and allow applications to access file and block data. That’s storage virtualization without compromise.
Once the ViPR Controller is implemented, a host of services can be added with ViPR Services which can be layered over both traditional and new storage and enable hybrid data types - all at cloud scale. ViPR Services extract more value from your existing investments by shifting away from array-specific services to holistically meet SLAs, mixed workloads, and Big Data applications. ViPR Services offers:
The ECS object service provides the ability to store, access and manipulate unstructured data such as images, video, audio, and online documents as objects on file-based storage such as VNX, Isilon and NetApp storage systems without having to rewrite or rework existing file-based applications. The ECS object data service is a software layer that works transparently with different hardware platforms.
The ECS HDFS data service gives organizations the ability to run analytics using well known industry Hadoop distributions on existing data stored across heterogeneous systems such as VNX, Isilon and NetApp arrays.
The ECS Block service enables organizations to support any application, file system, database, operating system or virtualization platform with a unilateral block storage consumption model across ViPR Controller-managed arrays.
With ViPR Controller customers can begin to realize true cloud value – such as on-demand storage-as-a-service and higher-level data services with the addition of ViPR Services.
Everything can be delivered as a service with ViPR Controller – storage services based on performance, availability and protection.
So why should companies consider undertaking the journey to Continuous Availability? The reason is that even with great business continuity plans and disaster recovery programs in place, organizations know that should a site be lost, it often will take weeks to restore full service, if at all.
Continuous Availability offers organizations the opportunity to:
Merge HA & DR resources into one availability solution
Eliminate downtime for a multitude of scenarios, including site loss, storage loss, infra loss, unscheduled downtime and scheduled downtime
Eliminate most DR scenarios
Eliminate RTO & RPO
5X9s of Availability across 2-metro distance sites
Potentially reduce compute costs 40-60%
Splitting existing clusters between sites
Using fractional provisioning
Leveraging existing technologies
Reducing operational complexity
Minimizing idle assets and OS maintenance
Now we will get into more detail on VPLEX & RecoverPoint integration with ViPR Controller.
Later, the SAP Application owner is ready to move the application into production. In this case they want to retain a copy of the application in the test & dev environment. For production, they need greater performance, availability and protection for the expected workload.
< CLICK >
They have a profile for SAP production already defined in ViPR Controller as well. Notice that the performance, availability and protection is more aggressive in this profile.
The SAP application owner provisions the production volume by selecting the appropriate profile through the ViPR Controller provided service catalog…
< CLICK >
And ViPR Controller automates the provisioning tasks to allocate the resources in Connectrix and VMAX for the medium volume from the storage pool that matches those characteristics.
We now have a medium volume that delivers the performance we need.
Lets take a closer look at how we deliver on the other requirements in the profile…
For our SAP environment we happen to have VPLEX and VMAX in two sites so that continuous availability can be provided across the sites using a distributed virtual volume.
< CLICK >
The profile definition calls for a mirrored pair to protect the data in two sites. ViPR Controller automates the process of provisioning the storage, sending the various commands to VPLEX to provision appropriate storage from the VPLEX managed pool of storage. Then ViPR Controller instructs VPLEX to create mirrored pairs between the two sites creating a distributed virtual volume. This virtual volume will ensure that if one array or even an entire site fails, the application data will continue to be available through the distributed virtual volume.
< CLICK >
Here we show the example of an Exchange environment that needs a similar capability, but it’s profile is designed to use a VPLEX to deliver a distributed volume between VMAX and VNX. In this case, ViPR Controller profiles can be created to specify the appropriate storage based on the performance level required and VPLEX can create the appropriate volumes across the storage in its pool of managed storage.
< CLICK >
A profile could also be defined to have VPLEX deliver a distributed volume using 3rd party storage arrays. Again, ViPR Controller can create the profile specifying the appropriate level of service that equates to the storage managed by the VPLEX and VPLEX provisions and creates continuously available volumes based on the commands sent from ViPR Controller which are driven by the profile selected from the service catalogue.
Let’s take a look at another common use case for ViPR Controller and VPLEX – Automating a tech re-fresh.
Here we show a medium volume that had previously been provisioned from an older array which has reached it’s end of life.
When a new replacement system is installed…
< CLICK >
ViPR Controller discovers the new array and storage pools through VPLEX configuring all of the settings required to make that storage part of the VPLEX managed storage. ViPR Controller then adds the newly available storage pools to the ViPR Controller virtual pool so that it is available for provisioning.
< CLICK >
ViPR Controller then instructs VPLEX to provision a mirrored pair onto the new storage so that the data from the old storage is mirrored between both arrays. Once the array data is mirrored between arrays…
< CLICK >
ViPR Controller automatically cleans up the old array and make it ready to be removed form the data center. All that is left is to roll the old array to the loading dock.
Lets get back to our Exchange application.
Our profile defined in ViPR Controller calls for protection to be provided with continuous data protection, or CDP.
Within site 1 we have the application running on VMAX, but we also have RecoverPoint installed along with a VNX storage array.
< CLICK >
When this Exchange Profile definition was selected in ViPR Controller from the service catalog, ViPR Controller automated protection that is delivered by RecoverPoint for the disaster recovery and operational recovery defined in the profile.
ViPR Controller now automates provisioning of VPLEX Local and VPLEX Metro volumes protected by RecoverPoint with XtremIO as the back-end storage in 5 steps, an 81% reduction compared to manual steps.
ViPR Controller can now automate protection of customers’ most critical applications with VPLEX MetroPoint topology using VPLEX Metro continuous availability and RecoverPoint CDP for both metro sites and remote replication for the third DR site. VPLEX MetroPoint can be configured in 5 steps, a 91% reduction.
With ViPR Controller, customers can take crash consistent snapshots for a set of VPLEX volumes within a consistency group leveraging VMAX or VNX or array-based snapshot technology.
It also allows customers who have VMAX and VNX arrays behind VPLEX, to take continuous copies for VPLEX distributed volumes on one or both legs. This allows administrators to protect their application data from a single failure at either high availability site. They can also detach the mirrored copy and export it.
ViPR SRM enables IT to address the operational challenges of today's multivendor storage environments. It provides end-to-end visualization, analysis, and reporting for VPLEX environments. It also works with ViPR Controller to establish a solid foundation for bridging the gap between your current IT infrastructure and the 3rd platform. Together, these offerings provide you with the tools you need to abstract, pool and automate storage resources for delivering a more agile infrastructure with lower and for predictable costs.
ViPR SRM enables you to:
Understand application to storage dependencies by mapping topology and relationships from the host, through the switch, VPLEX, and down to the LUN on the array.
Troubleshoot performance problems, analyze performance trends, and create performance reports for key stakeholders.
Create application chargeback or show back reports to better align the cost of storage services with business requirements.
Analyze capacity consumption and performance trends to improve utilization and optimize resources
Monitor compliance with configuration best practices and the EMC support matrix to ensure the environment is always configured right to meet service levels.
Provide line of business stakeholders with reports that demonstrate SLA compliance.
Monitor compliance with data replication policies for EMC replication technologies to enhanced data protection
Analyze capacity consumption trends to improve capacity planning and purchasing processes.
ViPR SRM helps you visualize application to storage dependencies for performance analysis and troubleshooting, understand capacity usage and trends for planning, validate configurations to ensure compliance to service levels and automate reporting. With greater visibility and insight, you’re armed with the information needed to optimize your VPLEX environment to improve return on investment.
ViPR SRM also offers an extensible framework to support investments in ViPR Controller.
In short, the best thing about the EMC Hybrid Cloud solution is that it enables your business and IT teams to come together to provide not just legacy or traditional IT services, but to provide services that translate into real use case for your end users. Whether they are End user services, like the ability to request and provision VDI desktops, File Shares, or even to kick off more complex onboarding workflows…
…to business services…that generally support traditional business applications and systems in your datacenter…like requesting block or file storage, configuring backup and recovery services for your virtual servers, or just configuring new virtual servers to load applications onto…
…to more complex next-generation application services, like PaaS systems, scalable analytics, or even deploying complex applications
It’s these SERVICES that drive business. They don’t care about the wiring, they just want the light to come on when they hit the switch…
Underneath those services, are the enabling services that allow us to build and support them…these “Cloud Enabling Services” expose all of the “CLOUD” attributes that we’ve been talking about as why you need cloud…
What makes this possible?
Best of breed infrastructure from EMC and VMware that leverages the market leading integration that exists between both sets of technology.
Which creates a totally seamless experience, both the user – the consumer of the services your IT organization provides – and you – where you get to leverage all of your existing knowledge of EMC and VMware technology – so you don’t have to learn anything new.
It protects your existing investments in hardware and knowledge – and extends them to seamlessly leverage public cloud services – like vCloud Air and other EMC Cloud Service Providers – so you can do even more for your business.
What you get is a fully engineered, interoperable, seamless hybrid cloud running on technology you are already familiar with providing a very wide range of services and applications – all from a single self-service portal.
You also get choice in the hardware infrastructure – deploy quickly with Vblock or VSPEX – or roll with the network, server and EMC and VMware technology you prefer.
Optional Depth:
Specifically, what’s in the solution?
EMC & VMware Technologies Used:
VMware: vCloud Automation Center, vCenter Orchestrator, vCenter Operations Manager, ITBM Standard, LogInsight, NSX or vCNS, vSphere 5.5, vCHS.
EMC: ViPR Controller, Storage Analytics, Data Protection Advisor, Avamar, RecoverPoint, VPLEX with custom integration workflows for ViPR, Avamar & DPA
Built On Vblock, VSPEX for accelerated deployment, or the customer’s choice of EMC infrastructure
Converged infrastructure brings together compute, storage, I/O and management into an integrated platform for the fastest deployment and assured performance. Only EMC data protection and availability solutions – Avamar, Data Domain systems, RecoverPoint and VPLEX - are certified and integrated into Vblock and VSPEX to ensure optimal data protection and availability.
Key messages:
VPLEX Metro HA configs add two-nines to whatever availability would be provided by underlying storage infrastructure
For example, VPLEX Cluster Systems and VNX storage arrays have five-9's. With VPLEX Metro HA configurations, the customer will realize seven-9's availability for the entire storage infrastructure.
VPLEX Metro HA leverages two separate VPLEX clusters. Therefore, availability is the sum of the availability of the two independent clusters. Five-9's plus 5-9's equals ten-9's in theory. In practice, we are measuring seven-9's.
Worldwide, we have ~1750 VPLEX clusters configured as VPLEX Metro systems with VPLEX Witness enabled. (Note each VPLEX Metro system has two independent clusters). All told, these have over 22 Million cluster run hours.
Top ten things MCC did that we were too busy to tell you:
1,000,000+ run hours of SAP and Oracle workloads
1000+ Oracle users, and 40,000 Oracle transactions / minute
100+ years of simulated failures
16 months of continuous global operations
10 component, board, infrastructure, operational failures every week
4 enterprise scale applications hosted with HA & CA
3 times storage elements than average North America customer
So why should companies consider undertaking the journey to Continuous Availability? The reason is that even with great business continuity plans and disaster recovery programs in place, organizations know that should a site be lost, it often will take weeks to restore full service, if at all.
Continuous Availability offers organizations the opportunity to:
Merge HA & DR resources into one availability solution
Eliminate downtime for a multitude of scenarios, including site loss, storage loss, infra loss, unscheduled downtime and scheduled downtime
Eliminate most DR scenarios
Eliminate RTO & RPO
5X9s of Availability across 2-metro distance sites
Potentially reduce compute costs 40-60%
Splitting existing clusters between sites
Using fractional provisioning
Leveraging existing technologies
Reducing operational complexity
Minimizing idle assets and OS maintenance