The Catalyst 9400 was not out when we were looking at new core switches. We thought about using a Nexus 9k but neither were as good of a fit as the 9500. Between price and performance it was a no-brainer for us to look at the 9500. We have not nor in the future would we …
Cisco Nexus Series Switches
Verified User
Consultant
Chose Cisco Nexus Series Switches
Improved software designed for high availability, performance in Data Center
The [Cisco Catalyst 9500 Series Switches] is very well suited for datacenter core and distribution models, they work very well as a replacement for the 3850 switches that many smaller organizations use for core switching. The Catalyst 9500 series switches also works very well as a top of rack aggregation switch for SFP and QSFP connectivity.
Fairly stable and easy to operate - but I haven't any experience with other brands, so I don't have anything to compare with. Fairly fast in the OS. Easy to upgrade - but with some SW issues. Sometimes clearing of log folders is needed. Access to bash is cool. Tips & cool tricks for operations could be nice to share.
Maintenance, upgrades, and software certification can be performed without service interruptions because of the modular nature of NX-OS and features such as In-Service Software Upgrade (ISSU) and the capability for processes to restart dynamically
FabricPath:
Enables each device to build an overall view of the topology; this is similar to other link state routing protocols. Each device in the FabricPath topology is identified by a switch-id. The Layer 2 forwarding tables are built based on reachability to each switch-id, not by the MAC address. Eliminates spanning-tree to maximize network bandwidth and flexibility in topological configurations, as well as simplify operational support and configuration. This enables a tremendous amount of flexibility on the topology because you can now build FabricPath topologies for Layer 2-based networks the same as for Layer 3-based networks
Overlay Transport Virtualization (OTV): Enables the Layer 2 extension between distributed data centers over any transport Layer 3 network
Well the ACI, so the AP, the controller give us good chances to run everything, to configure everything. But the monitoring itself, it's not that easy. You need other products like Nexus dashboard or something else to do the monitoring. This is a little bit of a con. It's another technology so everyone has to think a little bit of others to adapt this knowledge and it takes two years I think for a technician to understand what ACI is really doing and what we can do with it.
Implementing jumbo frames on interfaces of its fabric extender series (N2k, etc.) by editing the network QoS does not have to be a global configuration that would affect all its interfaces. It can be improved to become just an interface configuration.
Licensing on the NXOS is a bit complicated and expensive. I understand that the Nexus is made for core data center switching but it does not have to break the bank.
OTV technology is for Nexus only. Based on the advantage of the technology, it should be made vendor-neutral to accommodate other vendor devices.
Based on the site load and requirements, the switches meets our processing and fiber ports requirements. Also they have great potential for future programmability options we are looking into.
We will use it for a long time, since they are switches with great capacity and positive results, they promote the perfect operation of our data center, I always recommend them, because they are switches with good results.
In our environment (and especially during COVID and the Work From Home era), I would really like our network devices to be more resilient [against} errors in remote configuration. Having a standard, easy to use, configure, commit/deploy, rollback system should be installed, but it's not there. Other companies have been providing this for over 20 years, and Cisco does this on some of their devices. Why not all?
These switches are very fast. They've been designed to work within the data center. We connect them to Cisco UCS-B Mini servers with the storage being directly attached. They are able to handle the data traffic pretty easily. We can also move servers pretty fast from data center to data center without overloading them. This has allowed our company to stay running during any kind of conditional outage. We have come to really rely on them for business continuity.
Overall, Cisco has great products and I believe that they believe in the philosophy of a great customer experience. Although there have been a few technical support issues that caused a lot of company anxiety, in most cases, Cisco has gone above and beyond in making a valiant effort to help the customer solve any issues.
The Catalyst 9400 was not out when we were looking at new core switches. We thought about using a Nexus 9k but neither were as good of a fit as the 9500. Between price and performance it was a no-brainer for us to look at the 9500. We have not nor in the future would we consider other brands for core switching.
The Cisco 9000 stacks up quite well against the Cisco Catalyst 3850 switches. The additional features available in the Nexus 9000, such as VPN, FCoE, 40 gigabits, give us the ability to support the future needs of the company in our data center. The Nexus 9000 allowed us to condense our core and aggregation environment that comprised of 2 Catalyst 6504 and 2 Catalyst 6509 to a port of Nexus 9000. Although the Catalyst 3850 would be sufficient to handle routing, those features in the Nexus 9000 made it the clear choice for us.
The Nexus 3000 series switches are data center switches, so I would say they have similar security ability to other switches in this segment. I don't have a lot of experience doing more than basic ACL security on switches, but I know these can be integrated into other security solutions like Cisco ISE and 802.1x authentication. It could also be integrated into an ACI solution to add micro segmentation, which would bring in other security functions.
These switches did require a purchase of the DNA license on top of standard switch licensing, even if the functions are no needed in the environment.
The increased SFP & QSFP connectivity enabled for faster processing within our compute environment. While the system is too new to quantify the value added in this, it undeniably will help with its ROI.
