Arista Network Upgrade & Lifecycle Services

Arista platforms and services we cover

ENCSE is a Arista partner in India. That page covers the product portfolio and procurement; this one covers lifecycle services on equipment you already run.

Upgrading a data centre fabric

Arista estates are most often data centre fabrics, and the defining constraint of a fabric upgrade is that the redundancy which makes it safe is also what makes sequencing critical. Leaf and spine layers have to be advanced in an order that never removes more than one path at a time, MLAG peers must be taken individually rather than together, and the change has to account for the fact that the workloads above the fabric are frequently the least tolerant of packet loss in the whole environment.

The upside is that a correctly designed fabric upgraded in the correct order can absorb the change with minimal service impact. The precondition is that the redundancy genuinely works — and verifying that, rather than assuming it from the design document, is where our assessment focuses on these estates. A fabric with a quietly failed path has no headroom for a planned outage on another.

EOS also supports capabilities that can reduce upgrade impact substantially on supported platforms and version combinations. Establishing whether your specific hardware and target version support them is a planning question, since the answer changes what the maintenance window has to accommodate.

Where our Arista lifecycle work typically starts

• A data centre fabric has drifted several EOS versions behind and needs bringing current without disrupting the workloads above it.
• A security advisory applies to the running EOS train and exposure across the fabric needs establishing.
• Fabric redundancy needs verifying before an upgrade programme can safely proceed.
• Hardware is approaching end of support and a refresh needs planning against budget and data centre change windows.
• An estate has been inherited and its current software and lifecycle position is undocumented.

Frequently asked questions

Can you upgrade an Arista fabric without disrupting workloads?

With a correctly designed and genuinely healthy fabric, impact can be reduced to very little by sequencing so that no more than one path is removed at a time and MLAG peers are taken individually. The precondition is that redundancy actually works — a fabric already running on a degraded path has no headroom to absorb a planned outage elsewhere, which is why verification comes before the upgrade plan.

How do you handle MLAG pairs during an upgrade?

One peer at a time, with validation between — confirming the surviving peer is carrying the traffic correctly and that the upgraded peer has rejoined properly before proceeding to the second. Upgrading both together removes the redundancy the design depends on and turns a maintenance window into an outage.

Which EOS version should we target?

Generally a well-established release in a maintained train rather than the newest available, unless you specifically need a feature only present in a later version. Chasing the newest release imports change you did not ask for; the objective is usually to close a specific exposure with the minimum amount of other change alongside it.

Do you assess Arista estates as part of a wider health check?

Yes. Arista equipment is covered within our network device health check alongside whatever else is installed, with findings presented consistently rather than split into separate per-vendor reports.

Can you plan an Arista refresh?

Yes. Supply an inventory and we will map each device against published Arista lifecycle milestones, then produce a phased refresh roadmap that respects data centre change constraints and your budget cycle. Our Arista partner page covers the broader portfolio.

Related services

• Arista Partner India
• Data Center Networking
• Network Switch Upgrade
• Network Device Health Check
• Request an Upgrade Assessment