Top 5 questions to ask before moving towards hyper converged infrastructure

At EnCloudEn, there was an internal debate when I had proposed this topic as our next blog post. After all, HCI is our lifeblood and appearing to put forth an argument on why potential customers should not choose moving towards hyper-converged infrastructure didn’t initially sit well with the stakeholders.

But as we believe in the approach of educating customers on making the right decisions would actually result in happy customers. After all, in the long run, the decision to move towards hyper converged HCI has to result in significant augmented value.

As a starting point for this exercise, we went back to our customers to find out their achieved results vis-à-vis their initial purchase rationale, what has clicked and what not. We, in fact, went back to some of the other folks who did not move to hyper converged infrastructure. We understood their reasoning and the results that they had achieved with the HCI substitutes that they were using.

5 questions to ask before moving towards hyper converged infrastructure

So finally, without much more ado, let me present to you the top 5 questions that you as a driver of IT innovation in your organization should consider when you move to hyper converged infrastructure:

1. Are you really scaling?

The key point that everyone drives across as the primary reason to transition to HCI is flexibility in how you grow. So it’s important to understand how much and how fast you should grow to leverage the benefits of HCI. And it’s also essential to be able to bifurcate this growth in terms of your compute requirements and your storage requirements.

Traditional SAN based architecture will also allow you to grow your compute as and when you require like HCI. However, SAN based clusters are generally smaller than HCI clusters. Thus, your high availability cluster may run out of physical hosts sooner in a SAN than HCI.

In terms of storage too, SANs can have some amount of dynamic capacity addition in terms of adding enclosures. But that would usually involve some amount of pre-planning and ahead of time investment. It is in this area that HCI flourishes with its both vertical as well as horizontal scaling capacity.

So, if you are unsure of your growth pattern, you should definitely move to hyper converged infrastructure. Capacity addition decisions can be data driven with analytics derived from initial consumption trends on the HCI platform.

2. What’s your ROBO architecture?

While moving towards hyper converged infrastructure, You’ll heard a lot on how HCI enables a ROBO architecture. Let me put the myth straight – not true.

Most ROBO offices hardly have the capacity requirements of a full-fledged HCI cluster. And having a central HCI and a single node in the remote office isn’t the same.

Organizations that have gone ahead with an HCI cluster on their remote office, would actually need to make the financial decision driven by their own workload requirements. Not the additional benefits of HCI in a ROBO scenario.

Where things come in handy, is in having the same HCI platform in both the central and remote offices. This makes the replication of workload so much easier. With the same storage architecture on both sides, you can rely on your HCI vendor to be providing this secure replication without additional software.

Remember, you will need an HCI cluster in each office.

3. How high are your storage performance needs?

We have met several customers who as soon as we meet them start asking about the storage IOPs and benchmarking numbers. We have learned to identify such requirements into two brackets – organizations that would like to have great performance and organizations that have extremely high storage performance needs.

Being a private cloud OEM as well, it is comparatively easier for us to be neutral to this. Although our sales team do have a higher commission to make in case they are able to make an HCI sale over a SAN based private cloud, they are generally comfortable pitching either. In fact, we have customers who are so particular about their storage performance that they have carried out SAN based EnCloudEn private cloud deployments without even a SAN switch. They were so particular about their latencies that they preferred to omit that network hop.

Our rule of thumb is simple. Once customers get into the range of around a half a million IOPs or higher as per their requirements, we inevitably see the buying decision veer towards a SAN. After all, having a dedicated storage controller does have its benefits at the end of the day in such high-end scenarios.

It’s not that our IOPs are limited to such numbers, but its just that we want the customer to feel comfortable with his choice. In conclusion, most enterprise workloads would run brilliantly on HCI.

It is due to the above reasons storage needs make up as a huge factor to consider while moving towards hyper-converged infrastructure. But if you think that you have something that far exceeds this, you are free to make your decision.

4. What is your consolidation lifecycle?

As an organization, you would have a larger IT infra strategy in place. Over time, your team may have accumulated multiple disparate pieces of hardware and software to cater to your data center needs. It is time you asked yourself, what’s your broader outlook.

A key tenet of a hardware agnostic HCI vendor like ours is that you can deploy our software on any existing x86 servers. It can be a mix and match of both hardware OEMs as well as configuration. This kind of flexibility allows you to take stock of the kind of server farm that you may have built. You can identify which servers could be brought into a growing HCI cluster and their workloads seamlessly migrated.

So if you think that you can make an attempt to transition most of your datacenter workloads to an HCI cluster, please transition quickly towards that. But if you want to just run a 3-5 node HCI cluster as a silo on your datacenter, moving towards hyper converged infrastructure may not be the best decision for your team.

5. What’s your primary workload distribution?

HCI-VDI, a combination made for each other. That was where the whole inception of HCI started. But today, HCI clusters in datacenters have gone ahead to embrace a wide gamut of enterprise workloads. From web servers to mission critical apps, from databases to dynamic scale out VMs, we see almost all possible workloads today running perfectly on HCI platforms.

However, there are a couple of more insights that become important. For example, it’s not just raw performance, but also application compatibility that needs to be investigated. Like you may need confirmation from your application vendor whether the compression algorithm of the HCI is supported by them. Another example may be on how your HCI provides clustered file systems/volumes and whether the same is adequate for the clusters that you want to set up.

These are examples of finer details that a solution architect should work with you on. It becomes extremely important that you and your team verify and optimize for your primary workloads when you make your HCI buying decision.

All in all, we believe that though modern HCI solutions are easy to adopt and scale, it should still align with the strategic IT goals that you have set as a team. In this quest, we feel that these 5 questions can help bring clarity and a structured rationale to your decision-making process. 

Let’s together try and change the perception of IT in an organization from a cost center to a resource center.

Increase HCI Efficiency

Leave a Reply

Your email address will not be published. Required fields are marked *