How to Move to New Data Center and Avoid Losing Data?
Cloud services gain in popularity every year. In our estimates, some 70% of large Russian companies have used the cloud or other services based on commercial data centers, at least once. This consumption model has always been good for rapid data transfer from local to the provider’s infrastructure, enabling similarly fast service termination if it is not needed anymore.
However, user experience is often far from the ideal. There is a bunch of memes depicting a burning data center and captioned "Transferring our data center to the cloud". Why is there such a difference between the expectations and reality? Perhaps, it is due to an insufficiently qualified provider that cannot organize the migration correctly and avoid information system downtime, or lack of experience on customer side.
All migrations are different, especially if we talk about a large infrastructure with many business-critical systems. Data transfer period varies roughly from one week to two or three months. Depending on the complexity, the process cost may vary as well. For instance, customer may require minimum downtime, which actually means that all business apps are to be up and running during the entire migration term. Although some challenges can occur during the transfer process, missed deadlines or post-migration system ineffectiveness won’t be the case if you follow a few recommendations.
Prepare twice and migrate once
Preparing properly makes 90% of success. It is important to learn in advance the type of the infrastructure to be transferred to the cloud or data center, work out a migration plan that describes areas of responsibility (so-called RACI matrix) and considers system specifications and computing capacity needs. Sizing network infrastructure is also a must-do here. Remember that the channel bandwidth of both source and target environment affects the migration speed. Customers can carry out migration on their own subject to sufficient in-house competence, or, as they often do, engage a provider that conducts an express audit by interviewing the customer to make sure everything will be OK.
Always keep data protection in mind
Backup is a rule of thumb to address any emergency and prevent irretrievable data loss when moving infrastructure and systems from one environment to another. Moreover, it is advisable to take additional information security measures since, for some time during the migration, a part of the infrastructure will be in the target environment (i.e. the cloud), while the other part is still in the source one.
You may need app optimization, but it’s not a problem.
Business systems used in large companies are usually heterogeneous, with some of them being developed almost decades ago. Traditionally, such apps ran solely on physical infrastructure and, due to outdated architecture, are extremely inefficient in the cloud. It makes sense to leave such apps running on-premise or transfer as is to similar physical equipment in a commercial data center.
At the other extreme, there are cloud native apps. Designed for the cloud, such apps migrate from one platform to another without a hitch.
But most apps are somewhere in between; although quite new, they need a few adjustments to operate in the cloud. Our experience shows that it never hurts to do this. For instance, in case of an e-commerce app, infrastructure analysis before migration to the cloud may reveal bottlenecks slowing down website operation. Such optimization is a kind of inventory taking with all unneeded stuff removed prior to moving to a new apartment.
Embrace automation if possible
Automated infrastructure transfer tools can make life easier for both provider and customer. Today, such tools operate effectively even when changing a hypervisor (i.e. a special software to deploy virtual machines), while before data migration from the platforms with different virtualization means was considered an extremely complicated task. If automated migration is impossible, physical media and backups are used, thus significantly extending the migration period.
Fortunately, it’s a rare case primarily related to organizational limitations, not the technical ones. Here is an example from our practice: with a contract expiring soon, the provider did not want to lose the customer and therefore reduced channel bandwidth, hindering the automated information copying to CROC Cloud. So, exporting data on removable disks turned to be a rather tricky job for us.
The final stage of any migration is testing of how information systems operate in a new environment. Before that, it’s better to check everything once again and make sure that there is no risk to business. As a result, customers and employees continue to use the services they are accustomed to, while the company optimizes infrastructure costs or enjoys service of even higher quality.