The future of IT infrastructure lies in the cloud, and customers are increasingly choosing to move from on-premise to cloud solutions. The benefits are tremendous, investments are reasonable and risk posture is reducing with the presence of cloud-integrated security solutions. Thus you need a really strong reason for retaining the legacy infrastructure.
The Insentra train is all about creating great experiences and I would like to talk about one here. We recently onboarded a customer for their journey moving from Veritas Enterprise Vault to G Suite. Migrating to a cloud service was a big step and this was accompanied with various reservations, concerns etc. They had several requirements which are outlined below:
- Move all archives from Veritas Enterprise Vault to Google Vault – VFE licensed mailboxes
- The migration must be completed by a defined deadline
- The migrated data should be available for eDiscovery
- Full Chain of Custody must be maintained
Addressing most of these were not a challenge, but the requirement around VFE licensed mailboxes as well as defined deadlines got us thinking. To understand this better, lets dive further into the problem and our approach addressing the same:
- It is not possible to migrate data directly into a Google Vault – VFE licensed mailbox, unless using Google migration solutions (which does not support sources like Veritas Enterprise Vault). This meant that we required full mailbox functionality licenses and that would incur an additional cost. In order to mitigate the licensing cost, we processed users in batches and rotated licenses (described in the ‘Migration Process’ section below). Thus we could drive a large migration with a limited assigned set of licenses.
- Migrating data to Google is like a slow tortoise running, that makes it extremely challenging to predict and meet deadlines. Do we have a limit on the number of tortoise runners though? Hence we decided to benefit from parallelism and the results were amazing. We had 4 servers processing 30 mailboxes each in parallel and achieved an astonishing migration rate of approximately 5,000,000 messages per day. We could now calculate the time to completion, based on the number of servers, archives and messages.
Having said that, no engagement can be successful without synergy with the customer, and we were about to hit a major challenge here. To explain this better, I have described the tasks performed for each migration batch below:
Migration Process:
- We created migration batches (groups) by sorting archives from the largest to smallest. This meant that our initial migration batches (largest archives) took days to complete which was excellent use of parallelism.
- The next step was communicating each batch completion status to the customer, and they performed the below steps:
- Reassign the G Suite license from the current batch to the next one
- Assign VFE licenses to the batch that completed, thus preserving data for future eDiscovery requirements
We were about to hit the smaller batches soon, some which would complete within hours or minutes, thus making this a difficult process to sustain. How do we improvise? How could we make this more repeatable and predictable for our customer? We decided to reshuffle the remaining batches by equally distributing archives, thus ensuring that all future batches were of the same size and required approximately a day to complete. This hybrid approach helped us streamline processes with people.
In summary this migration was a unique blend of technology, processes and people. Our mantra to success was parallelism, repeatable processes and predictability, which marks another great experience created on the Insentra train.
