With my last post being a segue to more technical topics, I now have a promise to keep and must write something technical. Note the “must”, not “have to”.
If I were to pick a Cisco product that I am particularly enamoured with, our Unified Computing System (UCS) would be a top contender. Yes, I could talk about how only six years after entering the server market UCS blade-servers are now the top selling in both Americas. I could go on how the platform was developed from the ground up to offer the best CPU and RAM resource utilisation and efficient cooling. I could even tell you how UCS reduces the Total Cost of Ownership (TCO) of running your DC or branch compute, with its simple cabling and straightforward, single-pane management.
For me, however, UCS is personal. That is because it was the first Cisco technology I learned about in depth, back when I was still going through the CSAP graduate program. At the time, I was working in our Customer Proof of Concept labs (CPOC), and yearning to delve deeper into the Data Centre world.
That being said, I will do my best to keep this post objective, as hard as it can be. 🙂
This will not be an overview of UCS. Rather, it will be a quick update on the recent new developments pertaining to the platform. In other words, I am assuming the readers are already somewhat familiar with the following:
The many faces of UCS
So, on to the updates. With the release of UCS Manager 3.1(1), all of the aforementioned faces can be managed using a single firmware release – this signifies an important step of unifying the software trains for UCS B-series/C-series, UCS M-series and UCS Mini. There will also exist a 2.2(x) train to ensure legacy hardware support (think M1 blades and 1st generation of Fabric Interconnects), but you can expect all the new goodies to utilise 3.1(x).
What is so good about 3.1(1) apart from its feature parity across platforms is the fact it is HTML5-enabled. At last, I can stop telling my customers that an HTML5-based UCS Manager is coming, because it is already here. 🙂
Next on, the newly announced, 3rd Gen 40G Fabric Interconnects (FIs).
The two variants of the 3rd Gen FIs
A keen eye will immediately notice two differences between these and the 2nd Gen FIs – firstly, they (obviously) use 40G ports for the most part, and secondly, they come in fixed models (so, no expansion modules here). What is not so obvious is the fact that the two differ in their Fibre Channel (FC) capabilities. While both can handle IP-based storage protocols such as FCoE/iSCSI/NAS/SMB, only the top one (6332-16UP) has FC capabilities. The introduction of the 40G FIs also brings with it new Input/Output Modules (IOMs, also known as Fabric Extenders or FEXes) – the 2304, with 4x40G uplinks.
Please note that the above do not make 2nd Gen FIs obsolete – and there are currently no plans for announcing their EoS/EoL.
If there was one thing that UCS could not do in the past, it would be Graphics Processing Unit (GPU)-powered Virtual Desktop Environment (VDI) deployment in a blade-server environment. As we know, NVIDIA GPUs for rackmount-servers have been readily available for some time now, and deployed with great success on UCS C-series (with three times the attach rate of the competitors). And we could utilise UCS B-series for GPU emulation – but there were no physical GPUs available for these.
This changed with the announcement of new, mezzanine form-factor NVIDIA GPUs that can now be deployed with UCS B-series, supporting up to 16 users on a single blade. This is especially important for UCS Mini deployments, as VDI capability can now be added without the need to invest in rackmount servers, reducing both the cost and rack space used.
Last but certainly not least, the recently announced HyperFlex platform – Cisco’s cutting-edge entry into the hyperconverged infrastructure market – has UCS at its core.
Thus, Cisco became the only vendor on the market to cover the entire spectrum of Data Centre solutions ranging from computing and converged infrastructure, through hyperconverged infrastructure, to scale-out modular servers. Since our portfolio covers all these areas, our customers can rest assured we can propose a solution that best fits their needs, rather than seeing everything as nails while holding a hammer in hand – because with Cisco, you get the entire toolbox instead.
I hope this quick overview brought you up to speed with Cisco UCS. The platform has gained significant momentum since its release, and Cisco is constantly working on improving it even further. I am looking forward to future announcements, and I will be happy to highlight some of them in my future posts.
Meanwhile, coming up next: demystifying Cisco ONE!Tags: