Image courtesy of Shutterstock.
Microsoft SQL Server has become a business-critical database for a growing number of enterprises that rely on it to run a wide range of essential business processes. As enterprises look to continuously improve the efficiency of their data centers, they face the challenges involved in improving their ability to provide high availability and disaster protection for SQL Server.
A common strategy for providing high availability protection for SQL Server is to use AlwaysOn Availability Groups, a high availability feature included with SQL Server 2012 Enterprise Edition. It is positioned as an evolution of SQL Server Database Mirroring and an alternative to AlwaysOn Failover Clustering.
You can also use AlwaysOn Failover Clustering, which is included in both the SQL Server Enterprise and Standard Editions. While AlwaysOn Failover Clustering allows you to create a cluster in a physical server environment, it requires shared storage, which is not available in a cloud environment, and may not be practical in a virtual server environment. A third option is to add SANLess clustering software to AlwaysOn Failover Clustering. SANLess clustering provides high availability and more comprehensive data protection for a fraction of the cost of AlwaysOn Availability Groups which requires the very expensive Enterprise Edition of SQL Server
SANLess clustering software is an ingredient that enhances a Windows Server Failover Clustering (WSFC) environment by providing real-time, block-level replication to synchronize local attached storage. The resulting synchronized storage appears to WSFC as a virtual SAN, enabling you to create a SANLess cluster that eliminates the cost, complexity, and single point of failure risk of a SAN. SANLess clusters also help you save money and enhance availability in a variety of ways.
1. Use SQL Server Standard Edition with AlwaysOn Failover Clustering to Save Licensing Costs.
AlwaysOn Availability Groups requires SQL Server 2012 Enterprise Edition. SANLess clustering software lets you use AlwaysOn Failover Clustering, which is included in both Standard and Enterprise Editions of SQL Server as a more cost-efficient failover and disaster protection solution.
Figure1 shows a side-by-side comparison of software licensing costs for a traditional cluster using AlwaysOn Availability Groups to protect SQL Server Enterprise Edition versus a SANLess cluster using AlwayOn Failover Clustering and SANless clustering software (SIOS DataKeeper™ Cluster Edition) to protect SQL Server Standard Edition.
Costs are calculated for comparable two-node clusters with four, eight, and sixteen cores. Software Assurance licensing costs are also included. As shown, the SANLess cluster saves $13,124 in a four-core cluster, $33,448 in an eight-core cluster, and $74,096 in a 16-core cluster configuration. These savings include the purchase of SANLess clustering software, which is licensed per node. When used in multiple SQL Server environments, SANLess clusters can save several hundred thousand dollars in software licenses.
Figure 1 – Cost Comparison of Different High Availability Solutions
2. Reduce complexity and risk by protecting all of your databases.
Because SQL Server 2012 AlwaysOn Availability Groups only supports up to 10 Availability Groups and 100 databases per Availability Group (provided the hardware will support that many replication pairs), you might be forced to choose which databases to leave unprotected or to add significant complexity (and cost) to your configuration to create multiple clustering environments. Instead, use Windows Server Failover Clustering with SANless software that supports an unlimited number of databases.
3. Protect system databases from costly data loss and downtime.
SQL Server 2012 AlwaysOn Availability Groups only replicate user-defined databases, and not system databases (such as Master and MSDB). SQL Agent jobs and SQL Server logons are not automatically synchronized either, and will not fail over as part of the Availability Group. By using AlwaysOn Failover Clustering and adding SANless software, you can create a cluster that provides high availability protection for the entire SQL Server instance, including all of the system databases, SQL Agent jobs and account information, thereby ensuring complete recovery.
4. Cut management complexity in half.
With AlwaysOn Availability Groups you are managing two separate SQL server nodes, so whatever change you make to one node, you have to remember to make on the other. With SQL Server Failover Clustering instances built with SANLess clustering software, the nodes are automatically kept synchronized and identical. There is only one, highly available SQL instance you need to manage, and whatever changes made are automatically protected/failed-over.
5. Automate database administration.
Eliminating repetitive manual tasks and the risk of human error they pose can save thousands of dollars annually. SQL Server 2012 AlwaysOn Availability Groups are configured and managed at the database layer, not at the SQL Server instance. Therefore, administrators must reconfigure protection every time a database is added or dropped. SQL Server Failover Clustering with SANLess software protects the entire SQL Server instance and automatically includes databases as they are added (or dropped) in the protection scheme, saving time and reducing the risk of human error.
6. Get high availability and disaster recovery protection in a single solution.
An hour of downtime for business-critical applications, such as SQL Server, can cost tens of thousands of dollars. However, using a traditional shared-storage cluster for disaster protection can mean significant added costs and complexity. For example, you may need a second SAN at your disaster recovery site and/or identical server hardware at your source and target locations.
Instead, consider using SANLess clustering software that lets you deploy a multi-site cluster for high availability and disaster recovery without the need for a SAN at any site. You can configure it with all replicated storage, or a combination of shared and replicated storage.
7. Use the Cloud for Disaster Protection.
Enterprises that want to keep SQL Server and other important applications on-premises need a cost-efficient way to provide protection from on-site disasters. By using a SANLess cluster configuration, you can create a multi-node cluster with failover to a node in a cloud (e.g., Amazon Web Services EC2 or Microsoft Azure). This configuration gives you disaster protection without the need to build or lease an off-site DR location. SANLess clusters allow you to mix traditional SAN-based and SANLess clusters to provide both single-site on-premises and multi-site disaster recovery protection.
8. Protect all data and any data type with replication.
AlwaysOn Availability Groups only replicate SQL Server databases, leaving other data and data types unprotected. Critical or management data that may be associated with your application or system operations but resident outside of the SQL database is unprotected. SANLess clustering software with Windows Server Failover Clustering delivers complete protection for both SQL data and any data your application or IT operations rely on outside of the SQL database, including Microsoft Word, SharePoint, SQL Data, or any data that might be related to a custom application you need to protect.
9. Improve replication efficiency.
When an AlwaysOn Availability Group is configured in a Synchronous-Commit mode (required for high availability), it slows application write performance. SANless software such as SIOS DataKeeper uses efficient block-level replication that minimizes the performance impact of synchronous replication. SIOS DataKeeper supports both synchronous and asynchronous replication modes. Replication efficiency means you will need less costly hardware resources and deliver better service to your users.
10. Eliminate the SAN Cost and Single Point of Failure Risk.
A traditional SAN-based AlwaysOn Failover Cluster Instance adds risk as a potential single point of failure. SIOS DataKeeper Cluster Edition uses real-time, block-level replication to synchronize local attached storage, enabling you to create a SANLess cluster that eliminates the cost and complexity of implementing a SAN.
Based on SAN costs in a recent ESG Lab Report1, the savings shown in Figure 2 compare the TCO of two use cases: a small SAN for organizations requiring 112 TB of storage; and a large SAN for organizations requiring 218 TB of storage. SAN TCO includes hardware, support, power, cooling and labor. SAN software is not required in the small SAN model used. A SANless cluster can eliminate the need for a SAN, eliminating a five-year total cost of ownership (TCO) ranging from approximately $387,068 to $735,944 for mid-tier storage arrays.
Figure 2 – Savings Afforded by Using a SANLess Cluster
Note that a SANless cluster significantly reduces labor costs. A typical SAN requires dedicated support from both the vendor and either a full time employee or a consultant on retainer to perform maintenance and troubleshoot performance. By contrast, a SANless cluster requires no such expertise because it uses standard server hardware, WSFC, and low-cost, easy-to-configure, local storage. SANless software also provides an intuitive configuration wizard and a management console that makes it easy to install and manage.
1 Kaufmann, Aviv, and Mike Leone, “Dell EqualLogic PS Series Storage and TCO Analysis,” Enterprise Strategy Group Lab Report. January 2014. http://en.community.dell.com/techcenter/extras/m/white_papers/2043815/