The All-Flash Array (AFA) is Obsolete!

Posted on July 25, 2016 by Josh Odgers

Over the last few years, I’ve had numerous customers ask about how Nutanix can support bare metal workloads. Up until recently, I haven’t had an answer the customers have wanted to hear.

As a result, some customers have been stuck using their exisiting SAN or worse still being forced to go out and buy a new SAN.

As a result many customers who have wanted to use or have already deployed hyperconverged infrastructure (HCI) for all other workloads are stuck managing an all flash array silo to service some bare metal workloads.

In June at .NEXT 2016, Nutanix announced Acropolis Block Services (ABS) which now allows bare metal workloads to be serviced by new or existing Nutanix clusters.

As Nutanix has both hybrid (SSD+SATA) and all-flash nodes, customers can chose the right node type/s for their workloads and present the storage externally for bare metal workloads while also supporting Virtual Machines and Acropolis File Services (AFS) and containers.

So why would anyone buy an all-flash array? Let’s discuss a few scenarios.

Scenario 1: Bare metal workloads

Firstly, what applications even need bare metal these days? This is an important question to ask yourself. Challenge the requirement for bare metal and see if the justifications are still valid and if so, has anything changed which would allow virtualization of the applications. But this is a topic for another post.

If a customer only needs new infrastructure for bare metal workloads, deploying Nutanix and ABS means they can start small and scale as required. This avoids one of the major pitfalls of having to size a monolithic centralised, dual controller storage array.

While some AFA vendors can/do allow for non-disruptive controller upgrades, it’s still not a very attractive proposition, nor is it quick or easy. and reduces resiliency during the process as one of two controllers are offline. Nutanix on the other hand performs one click rolling upgrades which mean the largest the cluster, the lower the impact of an upgrade as it is performed one node at a time without disruption and can also be done without risk of a subsequent failure taking storage offline.

If the environment will only ever be used for bare metal workloads, no problem. Acropolis Block Services offers all the advantages of an All Flash Array, with far superior flexibility, scalability and simplicity.

Advantages:

Start small and scale granularly as required allowing customers to take advantage of newer CPU/RAM/Flash technologies more frequently
Scale performance and capacity by adding node/s
Scale capacity only with storage-only nodes (which come in all flash)
Automatically scale multi-pathing as the cluster expands
Solution can support future workloads including multiple hypervisors / VMs / file services & containers without creating a silo
You can use Hybrid nodes to save cost while delivering All Flash performance for workloads which require it by using VM flash pinning which ensures all data is stored in flash and can be specified on a per disk basis.
The same ability as an all flash array to only add compute nodes.

Disadvantages:

Your all-flash array vendor reps will hound you.

Scenario 2: Mixed workloads inc VMs and bare metal

As with scenario 1, deploying Nutanix and ABS means customers can start small and scale as required. This again avoids the major pitfall of having to size a monolithic centralised, dual controller storage array and eliminates the need for separate environments.

Virtual machines can run on compute+storage nodes while bare metal workloads can have storage presented by all nodes within the cluster, including storage-only nodes. For those who are concerned about (potential but unlikely) noisy neighbour situations, specific nodes can also be specified while maintaining all the advantages of Nutanix one-click, non-disruptive upgrades.

Advantages:

Start small and scale granularly as required allowing customers to take advantage of newer CPU/RAM/Flash technologies more frequently
Scale performance and capacity by adding node/s
Scale capacity only with storage-only nodes (which also come in all flash)
Automatically scale multi-pathing for bare metal workloads as the cluster expands
Solution can support future workloads including multiple Hypervisors / VMs / file services & containers without creating a silo.

Disadvantages:

Your All-Flash array vendor reps will hound you.

What are the remaining advantages of using an all flash array?

In all seriousness, I can’t think of any but for fun let’s cover a few areas you can expect all-flash array vendors to argue.

Performance

Ah the age old appendage measuring contest. I have written about this topic many times, including in one of my most popular posts “Peak performances vs Real world performance“.

The fact is, every storage product has limits, even all-flash arrays and Nutanix. The major difference is that Nutanix limits are per cluster rather than per Dual Controller Pair, and Nutanix can continue to scale the number of nodes in a cluster and continue to increase performance. So if ultimate performance is actually required, Nutanix can continue to scale to meet any performance/capacity requirements.

In fact, with ABS the limit for performance is not even at the cluster layer as multiple clusters can provide storage to the same bare metal server/s while maintaining single pane of glass management through PRISM Central.

I recently completed some testing with where I demonstrated the performance advantage of storage only nodes for virtual machines as well as how storage-only nodes improve performance for bare metal servers using Acropolis Block Services which I will be publishing results for in the near future.

Data Reduction

Nutanix has had support for deduplication, compression for a long time and introduced Erasure Coding (EC-X) mid 2015. Each of these technologies are supported when using Acropolis Block Services (ABS).

As a result, when comparing data reduction with all-flash array vendors, while the implementation of these data reduction technologies varies between vendors, they all achieves similar data reduction ratios when applied to the same dataset.

Beware of some vendors who include things like backups in their deduplication or data reduction ratios, this is very misleading and most vendors have the same capabilities. For more information on this see: Deduplication ratios – What should be included in the reported ratio?

Cost

Here we should think about what are the age old problems are with centralized shared storage (like AFAs)? Things like choosing the right controllers and the fact when you add more capacity to the storage, you’re not (or at least rarely) scaling the controller/s at the same time come to mind immediately.

With Nutanix and Acropolis Block Services you can start your All Flash solution with three nodes which means a low capital expenditure (CAPEX) and then scale either linearly (with the same node types) or non-linearly (with mixed types or storage only nodes) as you need to without having to rip and replace (e.g.: SAN controller head swaps).

Starting small and scaling as required also allows you to take advantage of newer technologies such as newer Intel chipsets and NVMe/3D XPoint to get better value for your money.

Starting small and scaling as required also minimizes – if not eliminates – the risk of oversizing and avoids unnecessary operational expenses (OPEX) such as rack space, power, cooling. This also reduces supporting infrastructure requirements such as networking.

Summary:

As shown below, the Nutanix Acropolis Distributed Storage Fabric (ADSF) can support almost any workload from VDI to mixed server workloads, file, block , big data, business critical applications such as SAP / Oracle / Exchange / SQL and bare metal workloads without creating silos with point solutions.

In addition to supporting all these workloads, Nutanix ADSF scalability both from a capacity/performance and resiliency perspective ensures customers can start small and scale when required to meet their exact business needs without the guesswork.

With these capabilities, the All-Flash array is obsolete.

I encourage everyone to share (constructively) your thoughts in the comments section.

Note: You must sign in to comment using WordPress, Facebook, LinkedIn or Twitter as Anonymous comments will not be approved,

Related Articles:

Things to consider when choosing infrastructure.
Scale out performance testing with Nutanix Storage Only Nodes
What’s .NEXT 2016 – Acropolis Block Services (ABS)
Scale out performance testing of bare metal workloads on Acropolis Block Services (Coming soon)
What’s .NEXT 2016 – Any node can be storage only
What’s .NEXT 2016 – All Flash Everywhere!

Peak Performance vs Real World Performance

Posted on April 17, 2015 by Josh Odgers

In this post I will be discussing Real World Performance of Storage solutions compared to peak performance. To make my point I will be using some car analogies which will hopefully assist in getting my point across.

Starting with the Bugatti Veyron Super Sport (below). This car has a W16 engine with 4 turbochargers and produces 1183BHP (~880kW) and has a top speed (peak performance) of 267MPH (431KPH).

The Veyron achieved the world record 267MPH at Volkswagen’s Ehra-Lessien test track in Germany. The test track has a 5.6 mile long straight. This is one of the very few places on earth where the Veyron can actually achieve its peak performance.

Now for the Veyron to achieve the 267MPH, not only do you need a 5.6 mile long straight, but the Veyron’s rear spoiler must NOT be deployed. Now rear spoilers provide down-force to keep stability so having the spoiler down means the car has a reduced ability to for example take corners.

In addition to requiring a 5.6 mile long straight, the rear spoiler being down, the Veyron can also only maintain its top speed (Peak performance) for 12 minutes before the Veyron’s 26.4-gallon fuel tank will be emptied, which is lucky because the Veyron’s specially designed tyres only last 15mins at >250MHP.

So in reality, while the Bugatti Veyron is one of (if not the fastest) production car in the world, even when you have all your ducks in a row, you can still only achieve its peak performance for a very short period of time (in this example <12 mins) and with several constraints such as reduced ability to corner (due to reduced aerodynamics from the spoiler being down).

Now what about Fuel Economy? The Veyron is rated as follows:

City Driving: 29 L/100 km; 9.6 mpg

Highway Driving: 17 L/100 km; 17 mpg

Top Speed: 78 L/100 km; 3.6 mpg

As you can see, vastly different figures depending on how the Veyron is being used.

There are numerous other factors which can limit the Veyron’s performance, such as weather. For example if the test track is wet, or has strong head winds, the Veyron would not be able to perform at its peak.

So while the Veyron can achieve the 267MPH, In the real world, its average (or Real World) performance will be much lower and will vary significantly from owner to owner.

At this stage you’re probably asking “What has this got to do with Storage”?

A Storage solution, be it a SAN/NAS or Hyper-Converged, all can be configured and benchmarked to achieve really impressive Peak Performance (IOPS) much like the Veyron.

But these “Peak Performance” numbers can rarely (if at all) be achieved with “Real World” workloads, especially over an extended duration.

To quote two great guys in the Storage industry (Vaughn Stewart & Chad Sakac):

Absolute performance more often than not, is NOT the only design consideration.

I couldn’t agree with this more. The storage vendors are to blame by advertising unrealistic IOPS numbers based on 4K 100% read and now customers expect the same number of IOPS from SQL or Oracle.

The MPG of the Veyron is like the number of IOPS a Storage array can achieve. It Depends on how the Car or Storage Array is used! The car will get higher MPG if used only on the highway just like a Storage Array will get higher IOPS if only used for one I/O profile.

As the IO size and profile of workloads like SQL & Oracle are vastly different than the peak performance benchmarks using 4K 100% Read IOPS, expecting the same IOPS number for the benchmark and SQL/Oracle is as unrealistic as expecting the Veyron to do 267MPH in heavy traffic.

But like I said, Its the storage vendors fault for failing to educate customers on real world performance so many customers have the impression that peak IOPS is a good measurement, and as a result customers regularly waste time comparing Peak Performance of Vendor A and Vendor B, instead of focusing on their requirements and Real World performance.

In the real world, (at least in the vast majority of cases) customers don’t have dedicated storage solutions for one application where peak performance can be achieved, let alone sustained for any meaningful length of time.

Customers generally run numerous mixed workloads on their storage solutions, everything from Active Directory, DNS , DHCP etc which has low capacity/IOPS requirements , Database, Email and Application servers which may have higher capacity/IOPS requirements to achieve and backup with are low IOPS but high capacity.

Each of these workloads have different IO profiles and depending on storage architecture may share storage controllers / SSDs / HDDs / storage networking all of which can result in congestion / contention which leads to reduced performance.

Before you start considering what vendors storage solution is best, you need to first understand (and document) your requirements along with a success criteria which you can validate storage solutions against.

If your requirements are for example:

Host 10TB of Exchange Mailboxes for 2000 users (~400 random Read/Write 32-64k IOPS)
Host 20TB Windows DFS solution
Host 50TB of Backups
Support 1TB active working set SQL Database
Host 10TB of misc low IO random workload
Have Per VM snapshot / backup / replication capabilities

Then there is no point having (or testing) a solution for 100k Random Read 4k IOPS, as your requirement may be less than 10K IOPS of varying sizes and profile.

Consider this:

If the storage solution/s your considering can achieve the 10K IOPS with the I/O profile of your workloads and can be easily scaled, then a solution able to achieve 20K IOPS day 1, is of little/no advantage to a solution which can achieve 12K IOPS since 10K IOPS is all that you need.

Now if your Constraints are:

12RU rack space
4kw Power
$200k

Anything that’s larger than 12RU, uses more than 4Kw of Power or costs more than $200k is not something you should spend your time looking at / benchmarking etc since its not something you can purchase.

So to quote Vaughn and Chad again, “Don’t perform Absurd Testing”.

In my opinion, customers should value their own time enough not to waste time doing a proof of concepts (PoCs) on multiple different products when in reality only 2 meet your requirements.

An example of Absurd testing would be taking a Toyota Corolla on a test drive to a drag strip and testing its 1/4 mile performance when you plan to use the car to pick-up the shopping and drop the kids off at school.

Its equally as Absurd to test 100% Random Read 4k IOPS or consider/test/compare a storage solutions <insert your favourite feature here> when its not required or applicable to your use case.

Summary:

Peak performance is rarely a significant factor for a storage solution.
Understand and document you’re storage requirements / constraints before considering products.
Create a viability/success criteria when considering storage which validates the solution meets you’re requirements within the constraints.
Do not waste time performing absurd testing of “Peak performance” or “features” which are not required/applicable.
Only conduct Proof of Concepts on solutions:
1. Where no evidence exists on the solutions capability for your use case/s.
2. Which fall within your constraints (Cost, Size , Power , Cooling etc).
3. Which on paper meet/exceed your requirements!
4. Where you have a documented PoC plan with a detailed success criteria!
As long as the solution your considering can quickly, easily and non-disruptively scale, there is no need to oversize day 1.
1. If the solution your considering CANT quickly, easily and non-disruptively scale, then its probably not worth considering.
The performance of a storage solution can be impacted by many factors such as compute, network and applications.
When Benchmarking, do so with tests which simulate the workload/s you plan to run, not “hero” style 100% read 4k (to achieve peak IOPS numbers) or 100% read 256k (to achieve high throughput numbers).

Scaling problems with traditional shared storage

Posted on October 24, 2013 by Josh Odgers

At VMware vForum Sydney this week I presented “Taking vSphere to the next level with converged infrastructure”.

Firstly, I wanted to thank everyone who attended the session, it was a great turnout and during the Q&A there were a ton of great questions.

One part of the presentation I got a lot of feedback on was when I spoke about Performance and Scaling and how this is a major issue with traditional shared storage.

So for those who couldn’t attend the session, I decided to create this post.

So lets start with a traditional environment with two VMware ESXi hosts, connected via FC or IP to a Storage array. In this example the storage controllers have a combined capability of 100K IOPS.

As we have two (2) ESXi hosts, if we divide the performance capabilities of the storage controllers between the two hosts we get 50K IOPS per node.

This is an example of what I have typically seen in customer sites, and day 1, and performance normally meets the customers requirements.

As environments tend to grow over time, the most common thing to expand is the compute layer, so the below shows what happens when a third ESXi host is added to the cluster, and connected to the SAN.

The 100K IOPS is now divided by 3, and each ESXi host now has 33K IOPS.

This isn’t really what customers expect when they add additional servers to an environment, but in reality, the storage performance is further divided between ESXi hosts and results in less IOPS per host in the best case scenario. Worst case scenario is the additional workloads on the third host create contention, and each host may have even less IOPS available to it.

But wait, there’s more!

What happens when we add a forth host? We further reduce the storage performance per ESXi host to 25K IOPS as shown below, which is HALF the original performance.

At this stage, the customers performance is generally significantly impacted, and there is no easy or cost effective resolution to the problem.

….. and when we add a fifth host? We continue to reduce the storage performance per ESXi host to 20K IOPS which is less than half its original performance.

So at this stage, some of you may be thinking, “yeah yeah, but I would also scale my storage by adding disk shelves.”

So lets add a disk shelf and see what happens.

We still only have 100K IOPS capable storage controllers, so we don’t get any additional IOPS to our ESXi hosts, the result of adding the additional disk shelf is REDUCED performance per GB!

Make sure when your looking at implementing, upgrading or replacing your storage solution that it can actually scale both performance (IOPS/throughput) AND capacity in a linear fashion,otherwise your environment will to some extent be impacted by what I have explained above. The only ways to avoid the above is to oversize your storage day 1, but even if you do this, over time your environment will appear to become slower (and your CAPEX will be very high).

Also, consider the scaling increments, as a solutions ability to scale should not require you to replace controllers or disks, or have a maximum number of controllers in the cluster. it also should scale in both small, medium and large increments depending on the requirements of the customer.

This is why I believe scale out shared nothing architecture will be the architecture of the future and it has already been proven by the likes of Google, Facebook and Twitter, and now brought to market by Nutanix.

Traditional storage, no matter how intelligent does not scale linearly or granularly enough. This results in complexity in architecture of storage solutions for environments which grow over time and lead to customers spending more money up front when the investment may not be realised for 2-5 years.

I’d prefer to be able to Start small with as little as 3 nodes, and scale one node at a time (regardless of node model ie: NX1000 , NX3000 , NX6000) to meet my customers requirements and never have to replace hardware just to get more performance or capacity.

Here is a summary of the Nutanix scaling capabilities, where you can scale Compute heavy, storage heavy or a mix of both as required.

CloudXC

By Josh Odgers – VMware Certified Design Expert (VCDX) #90

Tag Archives: pure storage

The All-Flash Array (AFA) is Obsolete!

Peak Performance vs Real World Performance

Scaling problems with traditional shared storage

Share this:

Share this:

Share this: