Around the Storage Block
Showing results for 
Search instead for 
Did you mean: 

Relax with 3PAR StoreServ & Survive That 3:00 am SQL Server Disaster Phone Call



Here's SQL Server management that let's you relax: Deploy your mission-critical SQL Server database on HPE 3PAR StoreServ and leverage value-added HPE 3PAR CLX, Adaptive Flash Cache and Priority Optimization.  It’s welcome news that brings a full night of sleep to your storage administrators.

HPE Storage MS SQL Server Management.jpgMission-critical databases like Microsoft SQL Server 2014 are deployed across a broad swath of industries, serving multiple use cases from Online Transaction Processing (OLTP) to Business Intelligence and Data Warehousing. Storage administrators pay special attention to their SQL Server SLAs and delivering 7x24 availability their holy grail.

The HPE Reference Architecture for Microsoft SQL Server 2014 on HPE 3PAR storage with Disaster Recovery will ensure that the mythical “3:00 am” phone call related to a SQL Server outage is no longer a source of trepidation.

This solution, virtualized by Hyper-V, is designed for geographic diversity, with SQL Server running in a local data center replicated at a distant disaster recovery location, for resilience against natural events such as earthquakes or power outages as well as outright equipment failures.

The one-minute phone call

Wait, just one minute? Yes, but before we get to that, let’s look at two key database resiliency related metrics on which a disaster-tolerant solution is designed and measured:

  • Recovery Point Objective (RPO) - RPO is generally defined as an amount of time prior to the disaster where all committed transactions can be recovered. A shorter RPO equates to a more resilient solution.
  • Recover Time Objective (RTO) - RTO is the amount of downtime that may elapse following a disaster before the database needs to be up and running and consuming new transactions.

Now back to the one-minute phone call. This HPE 3PAR solution is designed to achieve the following:

  • RPO of less than 30 seconds
  • RTO of about one minute (64 seconds measured in the creation of this reference architecture)

 So yes, in about one minute, the storage administrator can achieve resolution of an unplanned SQL Server 2014 outage. 

Please note that times may vary. There are several variables that can change the times such as the number of volumes, the distance between sites, the bandwidth of the links, and sun spots. 

Disaster recovery for SQL Server 2014 with HPE 3PAR Asynchronous Streaming

For disaster recovery at a geographically distant recovery site, Asynchronous Streaming isolates host server write IO performance from the latency effects of the inter-site link used for long-distance replication and from the remote array’s write IO service times. This is accomplished by queuing up data at the source and replicating to the destination. Host write transactions are written to the local volume and held in cache memory for replication to the destination volume.

Asynchronous Streaming provides crash-consistent, replicated copies of data to the other data center. Consistent data means that it’s “in order” as written by the host server. Consistency is essential for SQL Server to be able to recover the database after a system crash or disaster event.

In contrast, Synchronous Replication (where all I/Os to storage are synchronously mirrored to another storage system via synchronous remote copy) provides copies of the data with NO data loss, but host IO response times vary depending on the replication distance.

This distance induced network latency negatively impacts database transaction processing performance, especially in a flash storage deployment. In an all-flash world, you need a solution that keeps a “near exact” copy without impacting the service level expectations influenced by flash storage. As mentioned above, this Reference Architecture using Asynchronous Streaming, achieves an RPO of less than 30 seconds (results may vary).

Key solution elements

Let’s review the key elements of this solution as shown in the figure below.

3PAR StoreServ SQL ServerKey Solution Components DR.jpg


  • HPE 3PAR StoreServ 20800 storage array delivers the raw storage capacity, port scalability for complex disaster recovery solutions and large cache size to queue incoming IO data.
  • HPE 3PAR Cluster Extension helps to automate failover and failback between sites by integrating HPE 3PAR Remote Copy with Microsoft’s Windows Failover Cluster solution. HPE 3PAR Cluster Extension can rapidly reinstate critical applications at a remote site after an adverse event.
  • HPE SN1100E Gen 5 16Gb Fibre Channel adapter from Emulex delivers up to 1.2 million IOPS with a 2.5x improvement in latency at host when deployed in a 16Gb FC SAN. The SN1100E support the Virtual Fibre Channel (VFC) feature in Microsoft Windows Server.

Key findings and results of the HPE 3PAR SQL Server Disaster Recovery Solution

  • SQL Server with Asynchronous Streaming maintains fast local IO response times when deployed on a Gen 5 16Gb Fibre Channel SAN
  • Asynchronous Streaming provides RPOs of less than 30 seconds
  • SQL Server failover/failback, and failure recovery times are reduced with CLX
  • SQL Server cluster solutions integrate with Asynchronous Streaming and CLX
  • CLX automates the storage recovery process for lower RTOs

But wait there’s more...

HPE 3PAR delivers additional value capabilities, at no extra cost, especially suited for a hybrid flash HPE 3PAR StorServ 20800 deployment. These include:

  • HPE Adaptive Flash Cache (AFC) is a built-in array functionality of the HPE 3PAR StoreServ that uses capacity on solid-state drives (SSDs) (flash) to act as Level-2 read cache holding random read data that has been removed from the array’s DRAM read cache. This can be a cost-effective way of keeping more random read data on very fast media to improve overall system performance.
  • HPE 3PAR Priority Optimization (aka QoS) offers mixed and fluctuating SQL server workload environments, such as combining OLTP and Data Warehousing on a single host server can result in small block IOs waiting behind large block IOs. Applying a priority policy to the incoming I/O requests, users can maintain the quality of service (QoS) and performance level for a particular workload when competition for the array’s shared resources is occurring.

And what about that minute phone call?

The measured RTO with this solution was 64 seconds!

3PAR StoreServ SQL Server Data Recovery.jpg


Essentially, during a phone call taking a little over a minute, the disaster event would have been managed. This kind of smooth SQL Server management brings a welcome few more hours of sleep for your administrator!

Learn more about Microsoft SQL Server storage solutions.

Read another blog on Async Streaming: Data protection for flash – why Async Streaming is such good news



Meet Around the Storage Block blogger Dinara Alieva, Solution Marketing and Alliance Manager, HPE Storage.





0 Kudos
About the Author


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

Read for dates
HPE Webinars - 2019
Find out about this year's live broadcasts and on-demand webinars.
Read more
Read for dates
HPE at 2019 Technology Events
Learn about the technology events where Hewlett Packard Enterprise will have a presence in 2019.
Read more
View all