3PAR InfoSight

Discover News – InfoSight Comes to 3PAR and gets AI

HPE Discover is due to kick off in Madrid next week and HPE have continued their current tradition of releasing all the big news before the show.

I’ve been lucky enough to attend a few Discovers now and I have noticed some interesting trends in the type of announcements. Previously it was all about physical products, what were the newest models what were the hardware innovations in these products. This year all the key announcements are software based and focus on enhancing user experience.

InfoSight

Ironically for many years whilst storage systems stored lots of data on them, the data analytics for managing the systems was somewhat lacklustre. Data was gathered from these systems and sent back to base but it was a simple affair which alerted you to failed components such as a disk. Nimble Storage saw a unique opportunity in this area and acted upon it by developing their class leading InfoSight analytics platform.

Like other storage systems Nimble sends data home, but where things get different are in the breadth of statistics collected and what happens to them once they are sent back to base. Nimble realised that by taking the data from their entire install base and then using machine learning against this, insights could be developed to predict issues before they even occur. I have two impressive stats to back this up; 86% of issues are being detected and fixed before the customer even knows about it and 54% of issues resolved are outside of storage. This ability to diagnose issues outside of storage closed what Nimble termed the app data gap.

InfoSight adds AI

The first major announcement from HPE Discover is that InfoSight is now able to utilise AI. We discussed previously how Nimble had built a cloud analytics platform from all the data that was being sent home. InfoSight was able to analyse and predict future patterns using machine learning. The introduction of AI allows this to be taken to the next level by enabling more complex issues to be automatically addressed.

Recommendations will go beyond simple problem prevention to include resource optimisations, ensuring you getting the most from your system. Again this is not just for the storage itself but the whole ecosystem for example it may recommend the movement of a VM to a different volume. The name of this new AI functionality “AI recommendation engine” clearly indicates its future direction for further automation.

You can see more about the AI capabilities in the video below:

InfoSight comes to 3PAR

The second piece of InfoSight news is that it is coming to 3PAR. 3PAR has always had a rich amount of telemetry being sent home but up until now it has arguably not been fully utilised. But of course lots of data without the tools to interpret and act upon it is not much use and this is what InfoSight has enabled.

3PAR InfoSight

Like the Nimble system before it InfoSight on 3PAR will not just report on the storage itself but down to the application level. To benefit from InfoSight on 3PAR you need to be running 3PAR OS 3.3.1 have an active support contract and of course have your 3PAR configured to dial home.

Wrap-up

The Nimble AI recommendations and 3PAR InfoSight are both free for customers with a active support contract and are expected in January 2018

I’m sure that the HPE engineering team are working at a furious pace to get InfoSight out across to as many products, as quickly as possible. It’s clear to see that improved analytics would not only reduce operational issues but that the introduction of AI driven resource recommendations will drive the move to an automated data centre. It seems obvious that automated resource allocation recommendations would fit in nicely with Synergy where the ability to spin up and down workloads dynamically could then be orchestrated intelligently.

Looks like that acquisition of Nimble is paying off already.

Backup gets sexy – Nimble Secondary Flash Array

Since when did backups get sexy? It was only a couple of years ago the then VP of HPE Storage told us that backups were “like eating Broccoli”, “no one likes it although we have to do it because it’s good for us”. Well how things change in a couple of years with new kids on the block like Rubrik and Cohesity offering backup which have buzz around them. Well now HPE is getting in on the action with their shiny new Nimble Secondary Flash Array which is closely integrated with Veeam. The theme between all these vendors as we will see is getting your backups to do more. Not just backing up and forgetting about the data forever.

Hardware

Nimble secondry flash array hardware

Let’s start with the hardware and then work back from that. First thing to note is read the product name carefully Secondary Flash Array is actually a hybrid system taking Nimble back to their roots as a hybrid system, in which they used the CASL system which aimed to give close to flash performance with a mix of SSD and spinning disk. All SF Series models consist of up to 21 HDD drives and 3 DFCs (holding up to 6 SSDs). The table below shows the hardware specs for each model. Table comparing Nimble secondary flash array models

Today there are two models available the SF100 and SF300, you can see their full specifications in HPE Quickspecs. Both models are available with iSCSI and FC connectivity, the numbers appear to denote the rounded down/up raw capacity for the SF100 and SF300 respectively. Although HPE are claiming an 8:1 data reduction giving the SF100 for example a potential deduped capacity of 800TB

Veeam integration

Veeam logo

The integration with Veeam is quite simply that Veeam can talk to the storage so that it can leverage storage based snapshots. There are several advantages to this approach first of all you are offloading the heavy lifting to the storage, this removes the issue of VM stun and means that you take very regular snapshots. This allows for a small RPO something which is becoming increasingly important with threats like ransomware.

By integrating with Veeam this allows application consistent snapshots to be taken at a per VM level. This allows for more granular recovery and for a better level of assurance v crash consistent recovery. Those familiar already with Veeam will be aware of their Veeam Explorer tool this effectively cracks open backups to see their contents and allow granular recovery. This has been used previously with other storage vendors and for recovering application items like SQL.

Use cases

This is great but still why do I need a fancy smanchy box for my backups can I not just have some dumb disks? Well you could do that, but the vison for this Veeam and Nimble combo is to bring more power and flexibility to your backup solution, so that you can harness the backups. You could for example use it for standing up your dev and test. Rather than having the cost and complexity of maintaining a separate environment this will allow you to spin up VM’s from backup and isolate them using Veeam labs

The extra zip in performance, would also be of use for the Veeam instant recovery technology where you can spin up a VM’s direct from backups. Plus given the additional performance there is no excuse for not testing those backups using the Virtual Lab functionality. You can read more on the Nimble SFA at the official secondary flash array page.

Related

Getting the most from Veeam

Veeam Upgrade Steps

Veeam Version Numbers

Deleting jobs in Veeam

 

Thoughts on HPE’s Nimble Acquisition

It’s only a couple of months ago I was writing about the Simplivity acquisition and it seems HPE have been splashing the cash again, this time announcing the intention to purchase Nimble Storage.  I must say I really didn’t see this coming, so let’s take  a look at what Nimble are all about and what we might expect.

 Nimble

Founded in 2007, Nimble were one of the new breed of storage companies that grew up around flash.    They began offering hybrid arrays, later moving into all flash and recently announcing cloud based storage.

Tech start-ups are generally funded by investors whom seek a return on their investment, the two most common exit routes are an IPO or to get purchased. Nimble went down the IPO route and went public in 2013. Although it has not been a very successful flotation, as the company never regained its peak value seen in the first 6 months after going public. Hence we find ourselves in the somewhat less common scenario of going public and then being purchased. HPE have agreed to pay $12.5 per share giving a total valuation of $1.1 billion, second time lucky for Nimble.

 Tech

Unlike all flash Vendors such as SolidFire and Pure, Nimble took a bet on the future being in hybrid systems that contained a mixture of flash and spinning disk. Nimble’s architecture contained a flash layer for the acceleration of performance and spinning disk for capacity.  Nimble’s secret sauce was CASL, which stood for Cache Accelerated Sequential Layout and aimed to buffer and then optimise writes ready for spinning disk .  I have to say at the time I thought it was really neat, why would you want lots of expensive flash when a hybrid system plus CASL combined performance and capacity. Unfortunately for Nimble the price of flash dropped quicker than expected, companies saw the benefit of all flash and other vendors managed to use data reduction techniques that made flash viable.

 Nimble reacted to this by announcing a range of all flash systems last year.   Their systems feature the full suite of data reduction technologies (thin provisioning, dedupe, compression) and they also have QoS. Data mobility between Nimble’s all flash and Hybrid offerings is enabled via Unified Flash Fabric.

Key to the Nimble offering has been Infosight, this tool collects a significant amount of analytics from the system to ensure high availability and optimum performance. Originally name Proactive Wellness, Nimble advises that Infosight detects and resolves 90% of issues without the customer even knowing about it.  I can only assume the name change to Infosight was to reflect a product evolving beyond error detection, into one offering capacity planning and the ability to extend the predictive analytics beyond the storage array into vSphere.

 Recently Nimble have announced Nimble Cloud Volumes (NCV) which allow block storage to be presented to the cloud.  This block storage can be utilised in AWS and Azure, allowing mobility between the clouds plus richer data services. A NCV lives on a Nimble array managed by Nimble in their datacentre and is managed via a web interface. NCV are currently in beta.

 Thoughts

There is definitely some overlap between the existing midrange 3PAR and the Nimble systems. The strategy that plays out from here is going to be really interesting, it would seem logical that they would wish to minimise overlapping product lines to reduce costs so that development efforts can be focused.

 The press release emphasises the services surrounding the Nimble ecosystem, Nimble has class leading monitoring and analytics with its Infosight system.

Maybe we need to stop thinking about specific product lines and how Nimble product X will match up against 3Par product Y, and start thinking about IP and feature sets. Taking the key features from this new look storage line up reveals some key capabilities/features:

·       3PAR – ASIC, Thin Provisioning and virtualised storage, Scalable active active design

·       Simplivity –  Class leading dedupe/compression enabled via the OmniStack Accelerator Card plus data mobility

·        Nimble –  Class leading analytics and now cloud storage platform

This is quite some line up. The potential to combine these offerings and produce best in class products is going to Keep HPE’s developers busy.

Layering on top of this new tech portfolio a tool such as Synergy, to consolidate, orchestrate and dynamically re-composition the environment would allow a cloud like experience ideal for a company with ambitions to be the leader in hybrid cloud, I think  I see where this is going ………….