Is your organization considering migrating its data centre? There are several different reasons why many companies are taking that step. If yours is, the first question may be: “How?”. In this blog post, we consider the pros and cons of five such migration strategies.
Forklift Existing Equipment
Forklifting the existing equipment out of the old location and into the new one is a relatively simple strategy. It offers both a lower cost than many of the other options (as duplicate equipment is not required) and the peace of mind that comes from knowing there is a proven track record of the applications working on the equipment.
However, with the lower costs come several risks related specifically to the transportation itself: specifically, there may be a requirement for special security arrangements, there are risks related to the data and / or equipment being lost or ending up in the wrong hands, and there is also risk of problems being introduced during transportation. Additionally, the vendor’s ability to fix some of the older technology may be limited.
- Cannot pre-test new network performance in advance of migration (very high risk)
- Cannot pre-test if hardcoded IP addresses exist
- Significant business risk if problems are encountered during the first week or two after migration (no fallback capability)
- May not have a Disaster Recovery (DR) position/capability during the physical migration timeline
If you need to back out of this option once implemented, the choices are very limited as the time to relocate and reinstall the equipment back at the original data centre leaves virtually no recovery window in most cases. In order to mitigate the risk, a very detailed dry run / paper test before the equipment is moved is highly recommended.
Establish a Swing Environment and Physically Relocate Data and Applications
This option has the benefit of allowing the transfer of very large quantities of data in a relatively short timeframe;
- Ability to ‘shake down’ hardware in advance of migration
- Ability to pre-test network performance in advance of migration
- Enhances migration fallback capability and timeline
With this benefit also comes the risk of data loss during transportation or ending up in the wrong hands. There is also the possibility of incompatibilities between source and destination data centres and, as with the forklift option, there is a need for special security for the transportation of the data.
In some ways the risks are similar to the forklift strategy; however, what is different is that with this strategy, it is possible in the early stages to fall back to the old site. This requires careful planning and coordination with lines of business. Again, as with the forklift option, the choices to back out are very limited as the time to relocate and install leaves virtually no recovery window in most cases. To mitigate the risk, it is wise to go through a very detailed dry run / paper test before the equipment is moved.
Establish a Swing Environment and Use Data Synchronization Technology to Move Applications and Data
Data synchronization technology may include Peer to Peer Remote Copy (PPRC), or others. In many cases, this may be the best option, depending on the specific reasons for moving the data centre. It may be the best course of action if the data is synchronized between the two sites. With this strategy, organizations face less risk of data loss and security problems, as it utilizes a proven technology with a high success rate.
We have seen this deployed successfully in several scenarios, for example, during a major online application migration in which unexpected network problems were encountered. The problem was quickly identified, and services were switched back to the originating site within 2 hours, without the lines of business experiencing an unplanned outage.
This option may require significant network upgrades and there is a risk of possible incompatibilities between the source and destination data centres.
Establish a Swing Environment and Use Tools to Move the Workload
This is a great strategy if an application needs to be transitioned to the new environment quickly and tools may include options such as vMotion. The risks posed by security problems or data loss are lower than forklifting the equipment or physically transferring the data and applications.
As with the preceding strategy, it may require network upgrades and it doesn’t eliminate the risk presented by the possible incompatibilities between source and destination data centres. There is also the added risk that staff may lack sufficient experience with such technology. In the event there is a need to back out of this strategy, the utilization of automated tools can assist.
Canary Test for Select Critical Applications
Canary testing involves testing with a small group of users in the new environment. With a canary strategy, higher levels of support can be dedicated to those in the canary group to help identify any initial problems and the impact of a failure can be minimized as the move is limited to a specific set of applications.
Despite the benefit, fallback plans are still necessary for those users involved in the canary test. The fall back would be limited to the select group of users dependent on the swing environment established.
Customize Strategies to Meet Defined Needs
Whatever the reasons for moving the data centre workload, it is absolutely essential to ensure that the benefits, risks and costs associated with an effective and efficient move are thoroughly considered and analyzed. As with any major project, careful and detailed planning is critical the overall success of the initiative. After all, you’ll want to ensure that the move achieves the desired results and doesn’t incur any unforeseen issues for the business.
In our experience, it is wise to consider a hybrid strategy that uses a mix of approaches based on the business risk of failure, fall back options, costs, staff capabilities and other IT or business initiatives.