|
Six Secrets to SQL Server Availability |
|
|
Microsoft SQL Server has become the data management tool of choice for a wide range of business critical systems, from electronic commerce to online banking to just-in-time manufacturing to streaming media. There are many expensive, complex technologies that promise availability for SQL. Fortunately there are also simple, automated ways to get the highest levels of protection. The following six secrets to affordable SQL availability will help you to implement a SQL environment with little or downtime, zero data loss, and no added IT complexity .ONE: PREVENTING IS BETTER THAN RECOVERING You want to keep your SQL applications running continuously and avoid losing any data. However, most availability technologies work like airbags in a car. They are designed to mitigate data loss and minimize recovery time after the fact. The failure still happens. SQL downtime still occurs and some data—particularly data that is in transit at the time of failure—is lost. Examples of these technologies include tape backup and restore, replication, hot stand-by, and even clusters. A better solution focuses on prevention of failures. Consider technology that enables the SQL application to operate continuously through a failure event so there is nothing to recover from and no data loss. This technology can even allow you to repair and restore failed server components or network connections while continuously online .TWO: CLUSTERS ARE NOT A LOW-COST SQL AVAILABILITY SOLUTION Although cluster software is available for free or at low cost, the cost of a cluster system can be 40-50% higher than other technologies. Clusters require you to use the higher cost “cluster-aware” version of SQL and, in many cases, more costly SAN configurations. You may even need to purchase additional hardware components. And if you are running a custom application on top of SQL, you will likely have to pay for custom scripting to enable that application to work after a cluster failover process. THREE: YOU DON’T HAVE TO CHOOSE BETWEEN DOWTIME AND DATA LOSS Optimally, you want to protect your SQL application and any application(s) running on it, your operating system, and all of your data. Most technologies are designed to protect either your applications or your data. Few protect your OS and fewer still protect all three. For example, data backup provides some data protection, but requires a full application restart and data recovery, meaning long Recovery Time Objective (RTO). Clusters move operations from a failed server to a stand-by server, possibly lessening, but not eliminating the interruption in application service. Applications in a cluster are still vulnerable to faulty I/O from device drivers and a wide range of other threats. Choose a software solution that keeps your OS running when a server fails, maintains application availability even when devices fail—without loss of application “state” and ensures all of your application data is protected. FOUR: TESTING, ONE, TWO, THREE… Gartner analysts stress the importance of routine testing of your IT infrastructure in preparation for a disaster. Unfortunately, with many technologies, that means a full restart—read downtime—for your SQL applications. Without this testing you have no way of anticipating scenarios that will render these technologies useless. In a common scenario, a Windows administrator updates or changes something on one cluster server without meticulously performing the same task on the other server. As a result, the cluster failover process does not engage at all in the event of a failover. FIVE: DON’T DISCOUNT THE HUMAN FACTORS Focus on two human factors: those that cause downtime and those that restore uptime and a protected environment. According to IDC, human error accounts for as much as 25% of IT downtime. Systems that provide automatic uptime protection eliminate your reliance on a small number of key employees to restore operation. Although some technologies have automatic failover capabilities, they require IT staff to perform extensive disaster scenario testing to keep them operational. Disaster tolerant technologies continue to operate though disasters, eliminating the need for human intervention and allowing repair and replacement of damaged components while continuously online. Restoring service is not just recovering data and restoring availability of the application. For traditional technologies, this may involve bringing systems back to their fully protected state. You may need to replace damaged server hardware, restore lost data, replacing redundant network connections, and/or reloading applications or an operating system. SIX: USE AUTOMATED SQL AVAILABILITY Marathon’s everRunTM software provides a completely automated path to SQL availability. Hundreds of organizations have deployed everRun to protect their SQL applications from hardware failures, software failures, and disasters. They’ve found us to be less expensive, simpler to operate and maintain, and more effective than any other availability solution they’ve evaluated. everRun completely synchronizes two standard Windows servers - including the OS, the SQL application, network interfaces, storage, and data - into a single operating environment with full hardware redundancy. Unlike cluster or failover technologies that require two fully configured systems, everRun creates a single Windows environment that operates like a standalone server. everRun software prevents interruptions and downtime by fully automating fault management, so that your SQL applications never see the failures. This design enables support for ANY SQL application without the need for customization or scripting. You simply install your SQL application and you’re done. Both everRun HA and everRun FT deliver the same automated capabilities, as well as a fully fault tolerant I/O subsystem. This means that failures at the I/O level, including network and storage adaptors, storage devices, etc. do not cause an interruption or restart of the application. The only difference between the everRun HA and FT configurations is that with everRun HA there is a brief interruption, if and only if, an entire server fails. In this case the system will automatically and immediately restart on the secondary server. everRun HA is ideal for the majority of SQL applications where a few minutes of downtime is acceptable, but data integrity and reliable recovery are critical. everRun FT is designed for your most mission critical SQL applications that simply can’t afford even a brief interruption in service. everRun FT prevents downtime even when a server fails, delivering true continuous availability without any interruptions. To sum up, everRun is ideal for SQL availability: Automated SQL application availability - Easy to operate and maintain. All fault handling and policy management are automated for you. Superior availability – We’re up. Always up. No loss of SQL application state. No loss of data. Affordability – Getting started is up to 36% less than other options. Administration and maintenance costs are up to 55% lower than they are with clusters. Non-intrusive – Works with standard x86 servers, no application or OS modifications required. Any SQL app – no cluster awareness or scripting required. Remote availability – Install our SplitSite option to geographically disperse your SQL servers.
|
|
