vRack migrations and the new challenges of infrastructure

vRack technology has been offered by OVH since 2009 to establish private connections between servers, whether they are hosted in one datacentre or several. Over the last 10 years or so, vRack has undergone considerable development, in order to keep adapting to the latest technological transformations and changing customer requirements. We are now in the process of deploying version 3.0, and sat down to take a look at this larger-scale migration with Stéphanie Logerot, Programme Director, and Sébastien Séjourné, Run Director.

Picture of Sebastien Sejourne and Stephanie Logerot
Sébastien Séjourné and Stéphanie Logerot

vRack has become a key service for anyone building a hybrid cloud infrastructure, which is now a growing trend*, as it responds to the very latest challenges in IT services. Stéphanie Logerot explains that “with artificial intelligence, the Internet of Things and big data, vRack needs to be in a position where it can take on these future challenges. This is why OVH decided to update it, and create version 3.0”.

Upgrading from one version to another may seem like a pretty common task. Indeed, the OVH teams are used to managing digital service developments, and upgrading from one version to another. This isn’t the first time this service has been upgraded, “but without a doubt, this is the most important upgrade from both a technological and organisational standpoint, as we want to carry it out as transparently as possible for our customers, and we want everyone using the old versions to enjoy the same benefits,” explains Sébastien.

With so many settings to take into account when an operation like this is carried out, we can clearly see the scale and logistical complexity involved. This will mean drastically changing the solution's architecture. Version 3.0 represents a significant transformation, particularly the move to a distributed infrastructure that offers more scalability and resilience. This upgrade needs to be carried out across all 28 OVH datacentres worldwide, as the vRack connects services that are hosted in multiple datacentres. Finally, since the change to vRack 3.0 is a network service upgrade, it involves a migration that must take place as quickly as possible, without causing any incidents.

Furthermore, it isn’t simply a case of moving from infrastructure A to infrastructure B — this migration affects an entire series of services, including Dedicated Servers, Private Cloud (vSphere) services, routing with vRack Connect, and our Load Balancers. Every service and specific process must be taken into account, as they will each require a new design. It’s also important to remember that this new architecture is based on hardware components, each with its own lifecycle. On top of this, we must take into account the diversity of the versions still used by our customers, some of whom employ unique, customised infrastructures. For this vRack upgrade, no less than eight different technical migrations have been identified and defined.

It is estimated that an operation of this scale will take several years to complete. The migration to vRack 3.0 started last year, with our Dedicated Servers, and since January 2018, a team of around 15 OVH employees have been working to design an efficient and transparent process for customers. Through reflection, sharing, software development, redeployment of throughput capacities, redundancy to improve fault tolerance, and careful organisation, the process is now well-established.

In early April, we were able to complete 30% of the work streams. By the end of this summer, many of our customers should have the new vRack version. Regular communication has taken place throughout this migration, in order to ensure a transparent process. We also regularly update the information regarding the different migrations on our status task website, so customers can follow their progress. You can view the status task here: http://travaux.ovh.net/?do=details&id=32595

Today, major providers of hyper-scale infrastructures are starting to confront the issue of managing industrial-scale hardware obsolescence. These problems are brand new, and they’re appearing at an unprecedented rate. In light of this, no single person can benefit solely from a specific specialism or expert feedback on this subject. The ability to face this issue is a major challenge for the entire community.

*Sources: https://www.bmc.com/blogs/cloud-growth-trends/: https://www.computerweekly.com/news/252438790/IaaS-emerges-as-fastest-growing-sector-of-the-global-public-cloud-market