Connect. Secure. Manage.
Cisco RoCE Network Design
Assess and design supported Cisco networks for RoCE workloads, coordinating endpoint settings, traffic isolation and congestion-control requirements.
Devices supported
across the UK
Client locations
supported
Traffic handled
at peak time
Average
response time
Average response is not a resolution time or a contractual guarantee. Support scope and response commitments are agreed for each engagement.
Cisco RoCE network design plans the Ethernet paths used by systems that transfer data through RDMA over Converged Ethernet. The right setup depends on the supported endpoints and workload. We work with the platform owners on traffic handling, isolation and tests. This checks the complete service, including what happens as links become busy.
Workload-led design
Start with application, compute and storage dependencies rather than the switch model.
Clear failure domains
Understand which links, devices and sites can fail together.
Operational clarity
Make fabrics, external routing and management responsibilities supportable.
Establish the supported application requirement
We ask the platform team to identify its adapter models, drivers, transport settings and documented network needs. RoCE is not one universal setup that can be copied between unrelated storage and compute setups. The supported combination of endpoints and Cisco switches must be established before changing production queues or flow-control policy.
The assessment distinguishes network duties from app tuning. We can design forwarding and traffic treatment. But the endpoint owner must confirm how the app uses RDMA and how it responds to congestion or path failure. This boundary avoids attributing every slow operation to the switches without evidence.
Standards & partnerships
Confidence in the team behind your network.
Cisco Partner
Cisco Meraki
Cyber Essentials
Cyber Essentials PlusMCI Network Services is operated by Proxar Ltd. Certifications apply to the certified legal entity and the scope of its certificates. View accreditation information
Trace the entire traffic path
We map the interfaces, VLANs, routing boundaries and any overlay involved in the service. RoCE versions and platform capabilities influence the permitted design, so assumptions about routing or encapsulation are checked clearly. End-to-end MTU also needs agreement across endpoints and every relevant forwarding segment.
A path diagram should identify where traffic is classified, queued and monitored. If a packet crosses a device with different marking or congestion behaviour, that transition must be understood. A consistent setup at the first and last switch does not prove consistency across the middle of the path.
A useful starting point
Not sure where to start?
Bring the requirement. We will help you define the engineering work.
Plan traffic classification deliberately
The design defines which traffic belongs in the RDMA class and how that classification is preserved. Trust boundaries matter: accepting arbitrary endpoint markings can allow unrelated traffic to enter a class intended for a controlled workload. We review classification at ingress and check the resulting treatment on each relevant hop.
Other traffic still needs a sensible policy. Management, app access and storage-control communications may share parts of the infrastructure without sharing the same needs. We avoid placing everything into a single protected class, which makes it harder to contain congestion and understand its effect on individual services.
Different environments. The same care.
Built around the work
your network supports.
Finance, banking and the wider private sector. We start with the systems and people that depend on your connections.
Finance & banking
Market-data paths, routing and provider coordination on clients’ behalf.
Financial-market services ↗Offices & multi-site teams
Reliable access between people, workplaces and business applications.
Office network design ↗Data centres & cloud
Network boundaries, workload connections and operational ownership.
Data centre networks ↗Coordinate flow control and congestion feedback
Priority flow control can pause chosen Ethernet priorities, while ECN can signal congestion to capable endpoints. Their roles and interactions depend on the supported rollout. We check the platform guidance, queue resources and endpoint response before proposing thresholds or enabling features.
The desired result is stable app behaviour under the agreed load, not simply the absence of a drop counter. Excessive pausing, prolonged congestion or unfair bandwidth distribution can still impair the service. We test these conditions and explain which settings are supported recommendations and which require workload-specific investigation.
Include failure and recovery behaviour
A resilient RoCE design must consider more than physical link redundancy. We review the path-selection method, available alternate capacity and the app’s recovery behaviour. The network may restore reachability while a job or storage operation still needs to retry or restart under the platform’s own rules.
Safe tests therefore involve the endpoint and app owners. We agree which link or device failures can be exercised, how workload effects will be measured and when the test will stop. A production setup carrying important storage traffic should not be subjected to exploratory fault injection without clear safeguards.
Validate using representative endpoints
Synthetic network tests can reveal physical or forwarding problems. But they do not replace validation of the actual RDMA workload. We correlate endpoint observations with Cisco interface, queue and congestion evidence. Testing includes relevant concurrency and data movement patterns rather than only a favourable isolated transfer.
The results identify software versions, adapter settings, topology and test conditions. This information is essential when a later driver upgrade or extra workload changes behaviour. A useful handover explains the normal operating range and the evidence to collect when performance departs from it.
Keep the design manageable
The operating documentation records traffic classes, marking policy, endpoint prerequisites, supported changes and escalation duties. Changes to MTU, queue allocation or transport settings should receive joint review because they can affect systems owned by different teams. Monitoring must make a degraded service visible before users report widespread disruption.
MCI can assess a current rollout or develop a new RoCE design as part of AI infrastructure networking or a storage project. Share the platform needs and a redacted topology first. We will define a networking scope and test programme without promising that a generic setup can deliver lossless operation under every possible condition.
Talk to a specialist
What does your network need next?
Tell us what you are planning, changing or trying to improve. We will help define a practical next step and the engineering scope.
+44 203 196 6566Choose option 1
sales@proxar.co.ukUK-wide service.
Most of our engineers are based in Kent, with a focus on London, Sussex, Essex and Berkshire.
Existing customers should use their agreed support and escalation channels. This form is for service enquiries, not a monitored emergency ticket.
Tell us what you need.
A short outline is enough. Please do not include passwords or sensitive network details.
Thank you. Your enquiry has been submitted.
Please check your name, email and message, then try again.
Your form session expired. Please try again.
Please wait a minute before sending another enquiry.
We could not submit your enquiry. Please call or email us instead.