Interestingly, the adjectives and superlatives in the title are not sufficient :-) Today, our HCI portfolio also comes in “smaller” and “more cost effective” and “much more flexible”.
Let me start with 3 “this I believe” strategic statements:
- I believe that the x86 server is becoming the base building block not just for compute – but through software defined stacks for storage, networking, and compute abstractions – the base building block for almost every workload. This is why my PowerEdge brothers and sisters approached the design of the 14th Generation PowerEdge servers to not only be an idealized compute and virtualization platform – but build in more than 150 improvements focused on making them be the ideal platform for SDS and HCI stacks
- I believe that HCI, and more generally SDS stacks are able to support the majority of x86 workloads by count – today. While there are workloads that for various data services, or CPU/persistence capacity ratio reasons are supported ideally on traditional storage models – they are not the majority by count. It’s important to realize we’re transitioning architectural approaches, and these transitions take time – but it’s not a tech thing at this point. This doesn’t mean “end of CI” or “end of external storage”. It does mean “more and more HCI/SDS every day”.
- I believe that there are approaches to HCI – a “pick software, pick hardware” approach (“Build”) or an integrated system approach (“Buy”) – both are real, both are material in the marketplace, both are important, both are non-transient. More on this one in another post coming soon.
The facts are pretty supportive:
Today’s news is another milestone on our journey around 1, 2, 3.
If you want to understand more – read on!
If you agree with statement #1, then lets ask: “what was designed into the 14th Generation PowerEdge platform which is germane when it comes to HCI and SDS models?”
Yes, of course, we start by taking advantage of the Intel® Xeon® Processor Scalable Family to provide significantly faster access to applications and data using much higher core counts. Each of those cores has a faster clock frequency, more memory channels, and faster memory. Those advances combine to provide up to 1.7x more processing power and 62 percent higher internal bandwidth.
There is improved support for SSDs and NVMe devices which is important in scale-out SDS deployments. Increasingly HCI all-flash approaches are becoming universal, and consistent performance for cache and capacity tier and pools is critical.
This is particularly true for HCI approaches where the cache tier plays a critical role and the slowest performing drive in a system can dictates the cluster level performance in some cases.
Even for approaches using SDS models like ScaleIO which are less dependent on a cache tier. Every storage stack has a different IO path. In cases like ScaleIO, the storage subsystem improvements represent something simple: a massive scaling improvement for massive pools of extremely low-latency IOps and petabytes of capacity. While today is about VxRail and XC where the cache tier is central to the IO path – our VxRack FLEX (which uses ScaleIO) HCI customers using customers can expect this 14G capability to come to them in the near future.
Beyond the storage subsystem design, in the 14th generation PowerEdge platforms there's also faster initialization and streamlined data storage management; new system boot approaches; increased drive cooling for improved reliability; expanded GPU capabilities, and, if needed, a common user interface for managing both Dell EMC server and HCI platforms.
If you agree with statement #2, then lets ask: “what about the HCI offers build on this “base building block”?
For customers who have standardized on VMware and are looking for the incredible unique capabilities of vSAN 6.6, there is nothing easier, nothing more turnkey, nothing more integrated (not only from an software engineering standpoint which does deep full stack lifecycle management, but also from a support standpoint) than VxRail.
This latest VxRail generation delivers performance for the most demanding applications with up to 2X more IOPS (input/output operations per second), and more than 2x faster response times. There also now 9x more predictable response times that can be sustained at sub-one millisecond – not only through the kernel-integrated capabilities of VMware vSAN, but also the incredible IO subsystem of PowerEdge.
vSAN Ready Node customers who are taking a software + add hardware approach (“Build”) have been experiencing this with the 14th Generation PowerEdge platform for a couple of months, now this capability is available for customers who want the full turnkey integrated system experience and want to never worry about integration, testing efforts ever again (“Buy”).
The new native VMware vSAN capabilities like Configuration Assist and VUM HBA tools are great – and represent the ongoing improvements to make the “Build” experience better. It’s great to see this continue to improve for customers who want to select software and hardware, then build, integrate and troubleshoot in the way that is familar to a vSphere administrator.
That said, the goal of VxRail is to to make all of these thing and much more all fade into the background, deliver an true integrated system experience – not only at one point, but for the life of the system. There’s a lot of software engineering and system level design/support that goes into this. There are a lot of people that work on this. It’s not for everyone, but its for people who want to get “out of the infrastructure business” (which means by definition that someone else is picking up the continuous integration work that a customer would otherwise need to do).
This is a short look at the updates we’ve done in just 6 months. For VxRail customers – each time, it was a “update now” click – with one group to hold accountable, period.
BTW – we’re pumped about VMware Cloud Foundation – which is great, and keeps getting better. VxRack SDDC is to VMware Cloud Foundation as VxRail is to vSphere/vSAN, and customers should expect us to continue to bring same the software + system engineering approach to VxRack SDDC in the near future. We are determined to bring these together – so customers can start with VxRail now, and step up to VxRack SDDC using VMware Cloud Foundation as needed.
Other things our VxRail customers are going to dig include the ability to perfectly right-sized deployments with millions more configurations spanning more processor options, new SATA SSDs, additional network connectivity options, and a 50 percent expansion of the GPUs that can be added – useful not only for extremely awesome VDI use cases, but also fascinating AI/Machine Learning use cases.
Historically, customer needed to choose the simplicity of HCI by trading off relatively rigid configurations and linear scaling models – not anymore. Not only is there more flexibility, but you can also scale in different ways – not just adding nodes, but adding components as needed:
VxRail has always had leading embedded Dell EMC data protection capabilities including Recoverpoint for VMs, we've now made available software-defined NAS support with Dell EMC Isilon SD Edge for use in remote offices or edge-to-core file deployments. Isilon SD Edge is included in VxRail in free and frictionless mode, and if you want to use it in full production use cases, it’s easily licensed.
And if that’s not enough – VxRail continues to be available in a simple, easy cloud model (opex only, pay as you go, prices decline over time) called CloudFlex:
When it comes to HCI we're just getting started. Soon there will be additional NVMe/SSD configurations that will be able to meet the needs of even the most demanding workloads – and we have much, much more in store.
Already the number of new workloads that don't benefit from HCI are more the exception than the general rule. Naturally, there are still some legacy applications with specific requirements for some data services, or other workloads that have very particular capacity scaling requirements (think HPC, think scaled/object storage). Those applications are well served by traditional converged infrastructure or specific designs.
But we’re at the point where HCI and SDS is the norm, the baseline – and if I were a customer, I would be asking “what applications are the exception, because I’m going to design for the general case”.
Modern HCI platforms such as VxRail on 14th Generation PowerEdge servers that are designed from the ground up to push the limits of scale to unimaginable new heights.
Thank you to the Dell EMC and VMware team that work so hard to bring VxRail to market – and thank you most of all to our partners and customers who put their faith in us!
Hi Chad,
It's exciting to see where HCI is going. We've been playing with ScaleIO and have been extremely impressed. I've been wondering about where storage is going and how it will impact HCI, particularly things like NVMEoF. VxRail/VxRack use TCP/IP...we've been hearing a lot about ROCE, iWARP, NVMe over FC, etc. I'm wondering how the HCI solutions will change as (if?) some of these technologies start to take off?
Posted by: Zonker | December 04, 2017 at 05:29 PM