How have major trends in hardware, such as converged infrastructure, wider usage of public cloud and the drive for greater consolidation, affected server demands?
As a data management company we at Asystec are witnessing first-hand the trend towards converged infrastructure and private/public cloud adoption.
The advent of virtualisation in reality lessened the reliance on branded server hardware that existed when talking about physical server workloads. We are focused on the applications and workloads and the server layer is just a vehicle for providing CPU, Memory and network connectivity. We find that our customers are less attached to the server vendors and are more willing to adopt commodity hardware/servers.
The demand for servers has not dissipated as a result converged systems and cloud offerings for a number of reasons. The advent of virtualisation brought with it the ease of creation of Virtual Machines/Virtual Servers (VM) in software often leading to VM sprawl meaning higher numbers of physical servers being procured to manage the ever growing virtual workloads.
There are ways to combat that this. It involves ensuring our customers have a VM Lifecycle Management (VMLM) policy in place and often Asystec works with our customers to automate these VMLM policies in software and in particular using VMware’s vRealize Suite.
While the benefits of converging the virtual machines (compute) workloads on the physical server are well known and common place the Data Centre Architecture designs the industry move to converge the network and storage layers at this physical compute layer has meant server demand is very much on the rise.
This is being done in software lending itself to the term Software Defined Data Centre (SDDC). VMware’s NSX solution looks to address the Software Defined Networking area with EMC ScaleIO software only solution addressing the Software Defined Storage layer.
These additional workloads on the physical servers is driving the growth of memory and CPU core speeds as well as CPU core density.
This (hyper) converged server layer with massive memory and CPU power with a scale out model is being positioned to have less latency issues or higher performance benefits than traversing the Storage and Ethernet networks to a central Storage Array(s). The total cost of ownership model needs to be calculated as not all workloads are equal and with that not all workloads will suit a particular architecture whether it be a Scale Up or Scale out design.
While this abstraction of the Data Centre into Software has meant the hardware platform has been viewed in some corners as commodity an engineered platform/solution over a build your own will reduce risk, cost and complexity.
As the topic in particular is server related Asystec have seen significant customer benefits with the likes of Cisco UCS servers can bring to a solution. UCS abstracts the entire identity of a physical server into a profile that can be migrated across hardware, in effect virtualising the entire server element of a private cloud.
Asystec are specialists (EMC Gold Partner and VCE Gold Partner) in deploying converged infrastructure such as VCE Vblock & VSPEX and the advantages UCS brings to these engineered solutions are a huge selling point for our clients.
We are seeing greater consolidation ratios in virtualised environments and cloud deployments but this doesn’t mean less servers are required.
The server market is vibrant and healthy but care must be taken in the Solution design and Operational management as the server count in a solution will be impacted as a result.