Around the Storage Block
1752805 Members
5576 Online
108789 Solutions
New Article
StorageExperts

HPE 3PAR Peer Persistence: Unattended Transparent Storage Availability

 

Peer Persistence enables HPE 3PAR StoreServ systems located at metropolitan distances to act as peers to each other, presenting a nearly continuous storage system to hosts and servers connected to them.  This post details Peer Persistence. 

Introduction to Peer Persistence 

3PAR Peer Persistence.jpgAll consolidation projects are essentially stacking as many eggs possible in a basket. So it is critical to manage availability by having plans to mitigate risks due to catastrophic infrastructure failures. Any infrastructure outage will have a wider impact in consolidation environment and recovery will be complex involving multiple applications. To mitigate against such risks it is critical to build a resilient infrastructure with sufficient redundancy. Over and above such an infrastructure, it is critical to have robust recovery automation to avoid manual operations, which are vulnerable to costly mistakes in tense situations.

At compute and network infrastructure level, clustering technologies built into Virtualization layer can detect and automatically respond to any application failures and host infrastructure failures in a robust fashion, eliminating error prone manual recovery procedures.  

At storage infrastructure level, it often requires using synchronous replication between two arrays for redundancy at different fault domains (or availability zones). In addition, to ensure the recovery automation from failures (failover) works throughout the entire infrastructure stack, it is required to coordinated failover actions across the entire infrastructure stack. This often requires complex integration software that integrate the host clustering’s software’s failover mechanism with the control of synchronous replication directionality at the array.

HPE 3PAR Peer Persistence provides a simpler solution which allows you to deploy a totally transparently highly available storage infrastructure so that there is no need to coordinate with the compute and network layers. Any failure in the storage infrastructure, be it an array failure or a total failover of a fault-domain (site), Peer Persistence will automatically redirect IOs to the prevailing storage thus ensuring transparent storage availability.

3PAR Peer Persistence How it Works

3PAR Peer Persistence deploys the following three key techniques to provide a transparently, highly available storage infrastructure:

  1. Synchronous Replication to replicate data between two arrays
  2. Same WWN for LUNs across two arrays with same data
  3. ALUA to control on which array a LUN is accessed at any given time

3PAR Remote Copy provides a robust synchronous replication mode. In synchronous mode all writes are acknowledged to the host only after the write IO is safely persisted in both the arrays. One are more 3PAR virtual volumes can be configured as a Remote Copy group and replicated synchronously between the arrays. More than one such Remote Copy groups can be configured. The Remote Copy groups can independently replicate data in either direction. This the keystone technique for Peer Persistence that provide two copies of data in two different arrays at different fault domain.

WWN is standards based identification mechanism used to identify objects in a SAN. In a SAN, each LUN is uniquely identified by a WWN associated to it by the array. In a redundant storage infrastructure a LUN is replicated to another LUN in a different array. If both the LUNs on the arrays have different WWN then the hosts will treat them as different LUNs even though they have the same data. A resilient storage infrastructure must handle failures transparent to the host and application to avoid any coordination. To achieve this the storage infrastructure must be able to dynamically assign the same WWN to any LUN with the exact copy of data on any array. This will make the changing from a LUN to a clone of it in a different array transparent to hosts and applications. 3PAR Peer Persistence implements a robust and clever algorithm to do this.

ALUA is an industry standard protocol for identifying optimized paths between a storage system and a host. The ALUA protocol allows initiators on the host side to coordinate with targets on the arrays side to choose and use the optimized path for a given LUN. 3PAR Peer Persistence uses the ALUA protocol to instruct the hosts to use the appropriate path to a LUN, based on which 3PAR array’s volume is currently serving it. The 3PAR Peer Persistence robust failover algorithm will make the paths Active to the 3PAR array that is serving the LUN and will make the paths to the other array to which the data is being replicated as Standby.

Peer Persistence in action

Here is a schematic diagram showing how a Peer Persistence is deployed.

HPE 3PAR Peer Persistence.pngHere are the salient aspects of the solution design that is required for a successful Peer Persistence deployment:

  1. There is sufficient redundancy and isolation between critical components of the solution as a single failure does not affect multiple components of the solution.
  2. The host access SAN is extended across the failure domains (availably zone or sites) so that there’s physical connectivity from all hosts to LUNs on both arrays. The SAN is extended over an owned or leased fiber link.
  3. There is a quorum server that is used as secondary mechanism to determine the status of the peer 3PAR array in a Peer Persistence setup. As this is a critical component it must be in its own availability zone, in a third site. It must be totally independent of both the 3PAR arrays. This is a simple process running in a small virtual machine. The Quorum Witness is required only when the links between the sites with the 3PAR arrays is not available. The Quorum Witness is not involved in the IO path, its failure does not affect normal IO operations on the Peer Persistence protected LUNS. However for a successful failover of Peer Persistence Groups, it is required that the Quorum Witness is up and running.

 You can deploy multiple Peer Persistence groups between two 3PAR arrays. Each group will replicate in the direction they are set to replicate and are managed independently. Whenever a 3PAR array becomes unavailable Peer Persistence running on the surviving 3PAR array will detect it. The surviving 3PAR will then automatically take over the serving Peer Persistence protected LUNs (accepting application IO) on the failed array, by using the replicated volumes. This means the replicated copy on the surviving 3PAR array will be made as primary using ALUA all the IOs to the LUN from the hosts will be directed to it.

Where to use Peer Persistence

Massively consolidated and virtualized environment benefit the most from 3PAR Peer Persistence, but it can be used with any application as it works transparent to operating systems and applications running on them.  Any application can reap the automatic , transparent storage failover benefits of Peer Persistence as long as the application is running in a supported server Operating Systems version listed in HPE 3PAR StoreServ – Peer Persistence Host OS Support Matrix available at HPE SPOCK.

What would it cost?

HPE is one of the early storage vendor to provide this advance capability in mid-range storage arrays without requiring additional hardware. Peer Persistence is available as part of HPE 3PAR All Inclusive Multi-System license and need not be purchased independently.  There is no need for additional software on the hosts to coordinate replication with host level clustering technologies.

HPE 3PAR Peer Persistence is already certified with both VMware Metro Storage Cluster (VMSC https://kb.vmware.com/kb/2055904 ) and Microsoft Stretched Failover Cluster.

Going Deeper

If you are not using 3PAR Peer Persistence, see it in action in this YouTube video. The video showcases how an Oracle workload is transparently protected by Peer Persistence.

Read through this whitepaper “Break the boundaries of storage for more details: 

Call your HPE representative or solution architect to evaluate deploying it in your data center.

3PAR Peer Persistence is the most used replication high availability solution with 3PAR customers. Through its transparent automation capabilities, peer persistence reduces maintenance effort across your infrastructure stack while effectively mitigating risks associated with recovering from catastrophic failures.

HPE 3PAR will continue to improve Peer Persistence technology to handle even more corner case failure scenarios not discussed in this article. Watch out for more innovation from 3PAR on this functionality in coming months.


Meet HPE Storage Tech insider blogger Guhendran Devendran, Worldwide Product Manager for 3PAR guhen_3PAR.jpgHigh Availability  

 

 

0 Kudos
About the Author

StorageExperts

Our team of Hewlett Packard Enterprise storage experts helps you to dive deep into relevant infrastructure topics.