Peak performance vs Real World – Exchange on Nutanix Acropolis Hypervisor (AHV)

I wrote a post in April 2015 titled “Peak Performance vs Real World Performance” which discusses how benchmarks are not realistic and the performance shown in benchmarks can rarely be reproduced with real workloads. It has been one of my most popular posts, and I have had overwhelmingly positive feedback, with only a select few still pushing unrealistic peak performance benchmarks as being of value to customers.

I thought I would whip up a post showing an example of benchmarks vs real world performance requirements using MS Exchange Jetstress on Nutanix.

The below is a screen shot from Nutanix PRISM HTML based GUI showing a Virtual Machines Read/Write IOPS , bandwidth and latency during a MS Exchange Jetstress benchmark.

JetstressAHV20160105

The screen shot shows ~4000 Read IOPS and ~4000 Write IOPS at a latency of 1.59ms.

But what does the above really tell us and what does it mean to a customer?

I’ve been quoted as saying “Benchmarks are of little value without context specific to customer requirements!” and I stand by this statement.

Let’s now look at an example of a real customers requirement:

The below is from the Exchange server role requirements calculator and it is a screen shot from the Role requirements tab which shows an estimate of the IOPS required for the Databases and Logs for a single Exchange instance.

ExchangeIOexample

It shows the required IOPS being 536 for the databases and 115 for the logs.

Note: The sizing calculator was for an environment supporting 20000 mailboxes across 3 mailbox servers. As such, the above IO requirements are for ~6666 users.

So now that we have done the MS Exchange solution sizing (shown above is just the storage performance requirements), we understand the requirement to be around 651 mixed Read/Write IOPS per mailbox VM. We can then take a benchmark such as Jetstress and validate that the solution has sufficient storage performance.

To require the ~8000 IOPS the Jetstress test showed, we would need to scale up each Exchange instances to support have a much larger number of users and have each user send/receive 500 emails per day to reach this requirement.

8kJetstressIOPS

But in scaling up each Exchange instance to reach the peak IOPS that even this 3 year old generation Nutanix node can deliver we would vastly exceed the compute sizing recommendations for Exchange 2013 (being 24vCPUs and 96GB RAM) as shown by the calculator below.

ScaleUpExchange

As we can see, for an Exchange instance to require those peak IOPS, we would have to size the Mailbox server VMs with more than 10x the recommended vCPUs (24) and 15x the RAM (96GB). This shows that peak IOPS which can be achieved are not relevant in the real world.

In fact, Exchange generally does not require more than 1000 IOPS. Typically its requires much less, as my earlier example shows. So peak performance numbers are of little/no value as they can’t (and more importantly don’t need to be) reproduced in the real world.

With a tool like Jetstress we can configure a precise Mailbox profiles and test only what you require. If the solution can produce more IOPS than what you need (such as in this example), that’s fine for headroom, but in this day and age where Nutanix allows you to quickly and easily scale (Compute/Storage performance & capacity), I recommend designing for what you need in the foreseeable future (by this I mean 6-12 months) and scale if/when required.

What a benchmark does help you understand is how much headroom a solution has over and above your requirements which can help choose a solution to support mixed workloads, BUT the benchmark would need to be re-ran concurrently with suitable benchmarks for all other applications you intend on mixing to see how the solution behaves with mixed workloads.

As such, single application peak performance benchmarks are almost never valuable (to customers) unless your planning to run application specific silos. I strongly recommend anyone considering implementing an application specific silo, read the following article: Enterprise Architecture & Avoiding tunnel vision.

And… if you’re planning to run application specific silos and/or scaling up workloads to the point they need crazy IOPS, then you’re increasing the size of your failure domains, CAPEX and OPEX which is only doing yourself (or your customer) a disservice. But that’s a topic for another day.

I hope this example shows how real world requirements and performance is vastly different to what a benchmark shows and why peak performance benchmarks should be taken with a grain of salt.

I’ve always said the focus should be on gathering requirements and delivering on business outcomes, not focusing on performance which is typically only a very small part of a solution that delivers a successful business outcome.

Summary:

When sizing an MS Exchange solution on Nutanix, IOPS is not a constraining factor even for large scale deployments. The most common constraining factor is the Microsoft recommended compute maximums being 24 vCPUs and 96GB RAM, which is the same constraint regardless of if you run on Nutanix, or any other virtual / physical platform.

Related Articles:

Enterprise Architecture & Avoiding tunnel vision.

Recently I have read a number of articles and had several conversations with architects and engineers across various specialities in the industry and I’m finding there is a growing trend of SMEs (Subject Matter Experts) having tunnel vision when it comes to architecting solutions for their customers.

What I mean by “Tunnel Vision” is that the architect only looks at what is right in front of him/her (e.g.: The current task/project) , and does not consider the implications of how the decisions being made for this task may impact the wider I.T infrastructure and customer from a commercial / operational perspective.

In my previous role I saw this all to often, and it was frustrating to know the solutions being designed and delivered to the customers were in some cases quite well designed when considered in isolation, but when taking into account the “Big Picture” (or what I would describe as the customers overall requirements) the solutions were adding unnecessary complexity, adding risk and increasing costs, when new solutions should be doing the exact opposite.

Lets start with an example;

Customer “ACME” need an enterprise messaging solution and have chosen Microsoft Exchange 2013 and have a requirement that there be no single points of failure in the environment.

Customer engages an Exchange SME who looks at the requirements for Exchange, he then points to a vendor best practice or reference architecture document and says “We’ll deploy Exchange on physical hardware, with JBOD & no shared storage and use Exchange Database Availability Groups for HA.”

The SME then attempts to justify his recommendation with “because its Microsoft’s Best practice” which most people still seem to blindly accept, but this is a story for another post.

In fairness to the SME, in isolation the decision/recommendation meets the customers messaging requirements, so what’s the problem?

If the customers had no existing I.T and the messaging system was going to be the only I.T infrastructure and they had no plans to run any other workloads, I would say the solution proposed could be a excellent solution, but how many customers only run messaging? In my experience, none.

So lets consider the customer has an existing Virtual environment, running Test/Dev, Production and Business Critical applications and adheres to a “Virtual First” policy.

The customer has already invested in virtualization & some form of shared storage (SAN/NAS/Web Scale) and has operational procedures and expertises in supporting and maintaining this environment.

If we were to add a new “silo” of physical servers, there are many disadvantages to the customer including but not limited too;

1. Additional operational documentation for new Physical environment.

2. New Backup & Disaster Recovery strategy / documentation.

3. Additional complexity managing / supporting a new Silo of infrastructure.

4. Reduced flexibility / scalability with physical servers vs virtual machines.

5. Increased downtime and/or impact in the event hardware failures.

6. Increased CAPEX due to having to size for future requirements due to scaling challenges with physical servers.

So what am I getting at?

The cost of deploying the MS Exchange solution on physical hardware could potentially be cheaper (CAPEX) Day 1 than virtualizing the new workload on the existing infrastructure (which likely needs to be scaled e.g.: Disk Shelves / Nodes) BUT would likely result overall higher TCO (Total Cost of Ownership) due to increased complexity & operational costs due to the creation of a new silo of resources.

Both a physical or virtual solution would likely meet/exceed the customers basic requirement to serve MS Exchange, but may have vastly different results in terms of the big picture.

Another example would be a customer has a legacy SAN which needs to be replaced and is causing issues for a large portion of the customers workloads, but the project being proposed is only to address the new Enterprise messaging requirements. In my opinion a good architect should consider the big picture and try to identify where projects can be combined (or a projects scope increased) to ensure a more cost effective yet better overall result for the customer.

If the architect only looked at Exchange and went Physical Servers w/ JBOD, there is zero chance of improvement for the rest of the infrastructure and the physical equipment for Exchange would likely be oversized and underutilized.

It will in many cases be much more economical to combine two or more projects, to enable the purchase of a new technology or infrastructure components and consolidate the workloads onto shared infrastructure rather than building two or more silo’s which add complexity to the environment, and will likely result in underutilized infrastructure and a solution which is inferior to what could have been achieved by combining the projects.

In conclusion, I hope that after reading this article, the next time you or your customers embark on a new project, that you as the Architect, Project Manager, or Engineer consider the big picture and not just the new requirement and ensure your customer/s get the best technical and business outcomes and avoid where possible the use of silos.