With all the spotlights on the big HPE 3PAR storage announcement some days ago, I notice that not everyone is aware of the release of a new storage whitepaper called HPE StoreVirtual VSA Ready Nodes. Which is IMHO also a big step forward for everyone busy with StoreVirtual VSA.

So what is this all about and why is it important?

This document is THE BIBLE for everyone that wants to start with StoreVirtual VSA. It contains a list of all the things to think about when designing, sizing and implementing the solution:

• Architectural overview
• Reference configurations
• Performance estimates
• Networking configuration
• Storage configuration
• Installation checklists for hardware, hypervisor, networking and storage
• Installation best practices

Unless you go for the HyperConverged way (where everything like StoreVirtual VSA is pre-installed and configured for you), this is the reference document for you.

In my job role I get often questions about sizing and installation of a DIY StoreVirtual VSA solution. Which RAID controller, is cache important, what about tiering, expected IOPS, do I need 10Gb, etc. All evident and good questions! But now with an answer in an official HPE document.

Reference configurations

HPE created 4 reference solutions based on expectations like storage size and performance:

• HPE Ready Nodes Small
• HPE Ready Nodes Medium
• HPE Ready Nodes Medium Hybrid (with SSD tiering)
• HPE Ready Nodes Large Hybrid (with SSD tiering)

Not going to write out all the details here, but you have a good mix of DL360 and DL380 based solutions with storage going from 5 SAS drives up to a hybrid setup with 18 SAS and 6 SSD drives per node.

You get a complete list of the tested configurations including BOM (Bill of Material) for available capacity and expected performance. I know that this is just a reference with 3 nodes, make the math yourself if you have 4 or 6 or even more nodes.

Further in the document you get all details about the installation following all best practices for hardware, hypervisor, networking, storage and the VSA itself. All in 1 document, I told you it is the bible…

Go get the StoreVirtual VSA Ready Nodes whitepaper here:

http://h20195.www2.hpe.com/V2/GetDocument.aspx?docname=a00000521ENW

Another important document is the StoreVirtual VSA Design and configuration guide. This document can be found here:

http://h20195.www2.hpe.com/V2/GetDocument.aspx?docname=4AA4-8440ENW

Be social and share!

Probably you all heard already about the recent acquisitions of Simplivity and Nimble by HPE. Many articles are already written about this topic (here, here and here) so I am not gonna repeat the old news.

However, there is 1 big question that remains (and where HPE is not (yet) showing its long-term vision yet) is: what’s next?

Is MSA dead? StoreVirtual maybe? Is Nimble replacing 3PAR? What about the software-defined storage stack in Simplivity versus StoreVirtual VSA?

One thing that HPE should avoid is creating a gigantic storage and HCI portfolio like Dell/EMC where you have plenty/too much choice for your needs (and unmanageable) or will they go for the Swiss army knife that can become unusable as well when not properly designed? My advice: keep it simple.

Look at the past

With the acquisition of LeftHand Networks in 2008, HPE shows that it knows how to integrate a newly acquired company successfully in its portfolio. Today the software defined storage platform StoreVirtual VSA is available in DIY server&storage solutions, is the core of the current HyperConverged platform HC250/380 and is also available inside the Composable Infrastructure with HPE Synergy.

Also the 3PAR portfolio got extended seriously since the acquisition in 2010. Initially it was a high-end platform becoming a leader in the midrange, high-end and all-flash market…

Besides the portfolio HPE did a good job as well in integrating these 2 platforms with their converged management tool OneView, which fits in the software defined Composable Infrastructure vision thanks to the use of open API’s to an extended partner ecosystem.

Simplivity

Both Simplivity as Nimble has unique features that HPE should use to add to strengthen further their portfolio.

The secret sauce of the Simplivity solution is the OmniStack Accelerator Card that provides inline hardware deduplication and compression on the I/O stack.

The current StoreVirtual VSA portfolio has an extended list of data services, even with data federation to the 3PAR platform, but so far there is still no dedupe and compression. IMHO this card will be an added value in the current HC380 platform. First announcement by Antonio Neri on CRN was the launch of OmniStack on the DL380 hardware as a first step.

On long term, I see further integration as well with 3PAR and Synergy.

Harder decision will be the choice for the software defined storage strategy. Both Simplivity as the StoreVirtual VSA have their strengths. The VSA announced recently data mobility with 3PAR with Peer Copy, and is also a core component in the Synergy stack.
Simplivity scales up to 32 nodes (VSA only 16) and has hardware acceleration. But is no true SAN since it can be used by the HC solution storing VM data only. StoreVirtual VSA is a true block based SAN iSCSI solution delivering more infrastructure flexibility..
I hope both will merge together combining their strengths.

On software side, I see the newly announced HC380 Operating Environment 2.0 merging with the Simplivity UI… Both have their strengths so I don’t see any of them disappearing. It’s just software, right?

Nimble

The Nimble acquisition will be another challenge for the HPE Storage team to tackle. HPE has already an extended portfolio with HPE MSA (entry price point), HPE StoreVirtual (software defined storage), HPE 3PAR (midrange enterprise) and XP7 (high-end enterprise). 3PAR is all-flash optimized, MSA and StoreVirtual partially.

The biggest advantage with this acquisition however for me is not the hardware side, but the software stack called InfoSight and the recently announced Nimble Cloud Volumes.

InfoSight is the predictive analytics platform from Nimble that provides the ability to monitor customer deployed infrastructure from the cloud, apply machine learning and predictive analytics to radically simplify operations and deliver a transformed support experience.

One of the key values of Nimble is the high customer satisfaction numbers. This is partially thanks to InfoSight that gives a complete view across the infrastructure from hardware up to the application level. So Nimble can act very fast wherever a problem is seen in the solution stack.

Nimble Cloud Volumes (NCV) allows customers to have block storage to be presented to the cloud.  At launch this block storage will be usable in AWS and Azure, allowing mobility between the cloud plus richer data services. A NCV lives on a Nimble array managed by Nimble in their datacenter and is managed via a web interface. Know that NCV is currently still in beta.

HPE recently announced their StoreOnce VSA supported in Azure for data protection to a third tier in the cloud, however so far there was no primary storage connector to the cloud. NCV will be a nice add-on to fill the cloud gap in the HPE storage portfolio.

Conclusion

I am a big fan of both acquisitions since they add unique features to the current HPE portfolio which weren’t there yet.

I see 2 challenges:

– Hardware integration within the current portfolio, where I see a huge overlap from Nimble with the MSA offering, Simplivity is quite clear moving everything to DL380 and removing the 3^rd party hardware.
I expect that the current data mobility across StoreVirtual and StoreServ (3PAR) will be extended to the other physical and virtual storage platforms by means of the recently announced Composable Data Fabric.

– Software integration where I see HPE OneView as the underlying infrastructure management tool for configuration management of the hardware layer, and InfoSight on top of that for predictive analytics and integration with the hypervisor and application layer.

These tools will act on the hardware layer where the HPE Simplivity OmniStack HyperConverged Operating Environment 2.0 (or whatever name it will become) will act on top of that infrastructure facing the customers/users of the underlying platform(s).

We’ll see if this picture is correct in a few years… 2017 will be a challenging year for HPE…

Be social and share!

One of the nice features of the new StoreVirtual 3200 system is the support for performance tiering to SSD. HPE calls this, similar to 3PAR, Adaptive Optimization. In this article I will describe the process how to enable Adaptive Optimization.

Know that for using this feature you will need the Advanced Data Services license and at least 4 SSD disks. Due to the active/active controller configuration 2 RAID 1 sets will be created. 6 SSD’s or more allow RAID5…

Setup

In my SV3200 I have 8 SAS drives configured in 2 RAID5 devices of 4 disks each in 1 Storage Pool. I don’t have the AO license activate yet either.

In the Disks section you will notice as well that 4 new disks are added.

When logging on to the console I notice instantly the 4 new disks displayed in yellow in the Disks donut stating that these 4 drives are uninitialized and so not in use…

First of all I will need to add the Advanced Data Services license to the system to avoid error messages later in the procedure. This is done through the typical website where you register most of all HPE Servers & Storage licenses: https://myenterpriselicense.hpe.com/

The license is delivered in an envelope with the system and contains an EON Entitlement Order Number.

This number needs to be entered on the Licensing Product Activation page:

After having validated all your registration information you will obtain the license key that you need to paste in the StoreVirtual Management Console SVMC.

On the Dashboard, you will see the confirmation that all features are activated.

Once the license is added, I can create new RAID devices with the newly added SSD’s.

In the Storage Systems window select Add RAID in the Actions button.

In the window that opens, SVMC will propose a default configuration based on HPE best practices.

Notice the Configure option. Here you can eventually override the HPE proposed configuration and choose another RAID configuration.

Since I only have 4 SSD disks, the only option for me is 2 RAID1 devices for the active/active dual controllers.

Click OK to start the configuration.

When the configuration task is finished you will see on the Storage screen that there will be 2 newly created RAID devices, and 12 configured disks (8 before).
There are as well 0 available disks to configure.

Work is not done yet. Per volume you can now enable Adaptive Optimization. From now on all ‘hot data’ will be placed on SSD storage.

This is the end of how activating Adaptive Optimization on StoreVirtual 3200 systems.

Be social and share!

First there was the CS240/242-HC for VMware with the 15 minute deployment time, then the CS250-HC for Microsoft Cloud Platform System and VMware, and now there is the next step towards the hyper converged world.

With the launch of the new HPE Hyper Converged 250 we see some interesting options appearing:

Now it is possible to get started with a 2-node configuration instead of 3 and initially even 4 nodes which made the price rather high to get started.

Further there are more hardware a-la carte options like 1Gb and 10Gb, capacity models with all SAS drives, hybrid models with multiple SSD sizes and now also an All-Flash option with 6 x 1,6TB SSD’s per node. Further can you choose between 1 x 10-core CPU (new) or 2 x 8 or 12-core CPU’s per node.

All systems will be deployed with StoreVirtual VSA 12.5 which means we can use the Quorum Witness functionality for 2-node systems and for remote and branch offices…

2-node configuration

The HC 250 starts now with 2-node configuration and allows for single node expansion.

This is an interesting option for smaller companies or companies with remote offices. Now we had to fall back to the good old Proliant DL series and build a similar solution based on StoreVirtual VSA technology. Now the same procedures and management tools can be used to get the same user experience.

Those remote offices can now also benefit from the StoreVirtual 12.5 features like Quorum Witness for automatic and transparent failover in 2-node solutions.
1 single central share can be used by several remote sites, which makes this a low-resource alternative for the FOM aka Failover Manager.

More hardware choices

On hardware level we see more options available on the Apollo 2000 Gen9 platform used.

First you have the choice in the hybrid model to have 2 x 400GB but now also 2 x 800GB SSD’s. And now there is also an All Flash option with 6 x 1,6TB SSD’s.

Further you have now, besides the existing 2 x 10Gb SFP+ based option also a 4 x 1Gb RJ45 card. Again handy for those remote offices.

Besides the dual 8-core or 12-core CPU option there is also now a single 10-core CPU available.

This gives for an appliance based solution quite some flexibility:

Of course the solution is still deployed with OneView Instant-On in 15 minutes, and management is done through vCenter with the pre-installed OneView4vCenter plugin…

And there is even more to come… Keep following my blog, cool times ahead!

Be social and share!

Call out to all my students in the past, and people who are interested in becoming HPE certified.

HPE MyLearning released on 1 March 2016 the new V2 version of the 3 levels of storage certifications being ATP, ASE and Master ASE.

All information on the new certification can be found here: http://www.hpe.com/info/certification

All certified people should log in into the MyLearning portal to register for access to the WBT’s and additional learning…

For the certified people there are upgrade exams available. Also HPE created so called Reference Library documents that holds all information important to focus on to get prepared for the upgrade exam.

Remember the V1 exam 3 years ago? I wrote an article about it to set the right expectations since quite some people failed the exam… Since they didn’t know about the Reference Library!

Well now HPE learned from this and decided to create 3 Reference Libraries specifically for ATP, ASE and Master ASE’s… Go check them out!

First class taught by me is already done (10 new Master ASE V2’s!), some other classes are scheduled at LAI in the Netherlands… More to come!

Are your ready to become a Master ASE Storage V2? I am, come join the club!  

Be social and share!

This week HPE announced its 3^rd offering in its Hyperconverged portfolio.
After the HC250 for VMware vSphere and the HC250 for Microsoft Cloud Platform System (both based on Apollo’s 2U chassis with up to 4 server blades) now there is a DL380 based solution called the HC380.

2 remarks returned quite often when offering the CS250 solution:

• Storage options are limited: there are SAS only, SSD tiering and even all-flash offerings, but 6 drives per server blade gave you a lot of CPU and memory with a limited amount of capacity in some situations.
• For true VDI workloads, there is no option to add professional graphic cards.

These 2 ‘complaints’ are now solved with the HC380.

By choosing another hardware platform (the well-known DL380 Gen9 series, worlds most sold server) gives HPE now the flexibility to offer 24 drives per node instead of 6, and adds extended graphics support with up to 2 Nvidia GRID cards per node.

This gives you a variety of configuration choice points:

Scalability stays similar, starting with 2 HC380 nodes growing up to 16 nodes in a cluster:

As mentioned the storage options are extended now giving you the flexibility to add up to 24 disks per node. This is always done in blocks of 8 disks, and must be homogenous throughout the cluster.

Various options are there like SAS only and hybrid with SSD tiering:

Software update: VM vending machine

However this is not a hardware only change. Also on software side there is a major change happening trying to make the life of the administrator even easier.

OK with the CS250 we could deploy the system in 15 minutes and with the OV4VC plug-in we could do all admin stuff in our well known vCenter client.

But now with the new HC380 Management UI HPE shows that it can be done even easier as before. I see options for my daughter starting to help me from now in in my company.

This new software stack will deliver a solution lifecycle management for the hardware (firmware, driver and software updates), monitoring but also the hypervisors running on the solution. The UI allows in 5 clicks the complete deployment of a VM.

My good friend Calvin Zito was one of the first to try this new UI and made a cool video post on it. Check it out here:

His full article on the HC380 can be found on his storage blog www.hpe.com/storage/blog

Other software enhancements are the possibility to integrate with CloudSystem 9.0. By adding this stack this solution will create a Cloud-in-a-box experience… Cloud-ready in less than 2 hours…

As soon as HPE permits I will try to get a hands-on on this new HC380 and will write in detail about the unboxing and installation experience.

Be social and share!

OK found my way back home after a fantastic week in Las Vegas at the annual HPE Discover event…

Since I was asked to teach twice the Master ASE Storage class to get +50 people certified with the highest level of storage certification (and most actually did pass the exam, congratulations!), I did not really have a lot of time during the event for blogging on all the news. That’s why I write this article just now…
There are always the typical announcements bigger/better/more:

Bigger SSD’s for 3PAR

Currently the largest drive available in a 3PAR system is/was 3,84TB. At Discover are not 1 but 2 new drives announced being 7.68TB and the huge 15,36TB SSD drive. It lowers the cost per GB even lower than a 10K SAS drive, somewhere below 1,2$/GB usable compared to 1,5$ for a SAS drive.

However some side notes on this.

In the past 5 years the density of performance HDD’s doubled. For SSD’s it is 38 times. Insane… How long will the SAS drive stay there? I won’t be impressed with the next 2,4TB SAS drive someday…

On top of that, with features like Deduplication you can even store more than 15,36TB on that drive… Not an option for SAS drives, SSD only feature…

We should not forget that these sizes are good for customers who require capacity. If you do the math you should need 64 x 1,92TB drives or 32 x 3,84TB drives for a certain capacity which you can cover now with 16 x 7,68TB or even 8 x 15,36TB drives. Since a lot of licensing is based on drive count in a 3PAR system, the larger capacity drives will be famous.

However! On performance side I still prefer the 64 SSD’s since you will get a higher performance count with 64 SSD’s compared to 8 SSD’s only. So it will be up to you to find the right spot between capacity and performance. Luckily HPE has the right tools for this (Ninjastars).

Adaptive Sparing 2.0

The next version of Adaptive Sparing was announced as well which increases the effective overprovisioning space even further from every SSD inside a 3PAR system. This is achieved by extending overprovisioning into user space that’s not currently being used by the array. If space is required for volumes, it is instantly made available to the system without performance impact I will write an article soon with more in-depth technical information in this.

RMC 3.0

Also a new version from Recovery Manager Central was announced. New version 3.0 enables now flexible replication to StoreOnce for Oracle and SAP/HANA. In the previous versions there was already support for VMware and SQL Server. Also on RMC there is a more in-depth article on its way…

Composable Data Fabric

However for me the biggest ‘announcement without products yet’ was the Composable Data Fabric.

This is today more a vision than a real product launch, but I like a lot the idea of one Data Fabric with 3PAR for the All-Flash Data Center and the Composable Data Fabric for the SW-Defined Data Center all managed by one tool being HPE OneView.

One interface, 1 feature set, any workload in a federated setup. All you need in 1 storage platform.

What I like the most is that HPE (finally) pushes their VSA in the picture where it should have been already for years. I am quite a fan of this product seen the various articles I have published already about it and all the customers here around in Belgium using VSA technology…

So there are no actual product launches yet, but what I notice is the StoreVirtual platform even more in the picture for structured but also unstructured and file data… This would mean file and object storage would come besides the block storage already there.

2016 will be a fun (storage) year, believe me!

Be social and share!

At least for 3,5 more years when I am writing this article…

HPE released the official customer notice with the EOL (End Of Life) announcement of the HPE StoreVirtual VSA software recently. The hardware versions of StoreVirtual P4000 and SV3200 were already EOL since some time.

The obsolescence date is set to 31 December 2019, meaning the last day you can purchase a license…
Second important date to remember is the EOSL (End Of Support Life) which is 3 years later: 31 December 2022.

After that, the VSA software will keep on being functional when you are using the perpetual licenses (so you don’t lose any data), however there is no support from HPE anymore.

Alternatives

HPE does not foresee a 1on1 replacement for the StoreVirtual VSA software, however there are some alternatives. I am assisting all my customers today with the migration towards HPE SimpliVity or Nimble solutions, depending on their needs…

I know Nimble is more a hardware solution and SimpliVity a hyper-converged all-in-one solution, however I see SimpliVity as the way to go, since there are a lot of similar features between the 2 solutions (minimum 2 nodes, 2 copies of data across the nodes in a cluster, stretched cluster options, etc). It does not replace entirely StoreVirtual VSA yet, however it has some other nice features like deduplication, compression and build-in data protection (restoring TB’s of data in a matter of minutes) that makes the solution very interesting to replace a StoreVirtual solution.

Hope HPE will make the SimpliVity solution even more flexible regarding deployment options. To be continued for sure. HPE can always contact me for my opinion on where SimpliVity should be going…

The link to the customer notice can be found here.

Meanwhile I keep my StoreVirtual VSA based volume online in my home-lab, meanwhile more than 10 years with 100% uptime. So it is my personal challenge now to go for at least 13 years! All details on how I did it can be found here.

Be social and share!

The post RIP StoreVirtual VSA, long live StoreVirtual VSA! appeared first on BITCON IT consultants.

I got a lot of messages and meeting requests since the official EOL announcement of HPE StoreVirtual VSA. That much that I want to write this article to clarify all options out there. It looks complicated, but it isn’t…

First define for yourself (or your customer) what you really need. High Availability or Disaster Recovery? RPO (Recovery Point Objective) of 0, or is a few minutes acceptable? A hyperconverged experience, or do you want a ‘physical’ storage solution in your server room to satisfy the server, storage and networking experts (aka traditional IT)? Data protection on infrastructure, server or application level? Questions enough to make the right decision.

HPE SimpliVity

When I meet a typical StoreVirtual VSA customer, I will lead always with HPE SimpliVity.
Why? Because there are a lot of similarities (more you can imagine) between the 2 products.

Although HPE never called the VSA a hyper converged product (missed opportunity), it actually is one… Server and storage combined in 1 appliance for a higher efficiency and lower TCO. Simplivity spreads blocks of data (after deduplication and compression) on 2 separate nodes in the cluster… What does Network RAID 10 from VSA? If you don’t know, read my (still popular) article explaining what NetworkRAID10 actually does… Yep , the same.

VSA delivers a shared storage platform to your server. What does the OmniStack Data Virtualization Platform? Indeed… VSA runs on VMware and Hyper-V, Simplivity too…

Now why would you prefer Simplivity? Well that is very simple… It does more compared to the VSA… It does deduplication and compression inline, guarantees performance, gives you 1 management interface, and gives you a better data protection. When I do a demo for a customer who suffered a CryptoLocker attack for instance, showing him his fileserver can be online again in 1 minute, believe me it is a no-brainer…

The strength of the Simplivity solution are the 5 above mentioned features in the solution you will get all at the same time. No choice to make. And verify that the other similar hyper converged solutions can do the same… Believe me, not all 5 of them together at the same time… Use cases enough to refer to.

Want more info? Try to attend on of my sessions on Simplivity and become a guru in positioning Simplivity…

HPE Nimble

Not all companies buy their server and storage infrastructure all at the same time. Most customers have various generations of servers implemented in a staged way… There it will be a bit harder to position Simplivity since you are buying an all-in-one solution replacing server and storage at the same time.

If you need specifically a storage solution to connect to their existing servers (which are not end of life/support yet), that is on top very efficient and can deliver performance and capacity in a scale-up and scale-out way, HPE Nimble is the way to go.

The Nimble portfolio consists of All-Flash (AF series) and Adaptive (HF series) models delivering all-flash performance, which is very simple to implement and maintain, delivering fully featured data services in a very efficient way by using as well inline deduplication and compression.

And now with the latest release of NimbleOS 5.1 it delivers synchronous replication as well. More information in other articles on my website.

Don’t forget to mention that Nimble also features HPE InfoSight delivering a cloud-based monitoring platform delivering AI-based cross-stack analytics. This platform will make the life easier of the IT team delivering predictive support. HPE support will call you instead of you calling them. Watch a demo of InfoSight and you will be convinced.

VMware vSAN

VMware vSAN is also a valid option when you compare features: change local storage into shared storage for your virtualized environment. When sizing, it is important to follow the HCL (Hardware Compatibility List) from VMware, but this is rather easy when using the predefined VSAN Ready Nodes from HPE. There is also a very easy to use VSAN Sizer available.

A bit tricky will be those implementations with 2 server rooms where you will need 6 nodes with VSAN to get the same resiliency compared to a 2-node VSA solution with a quorum arbiter. VMware VSAN supports 2-node deployments, but this is positioned for ROBO deployments only.

Veeam Backup & Replication

Veeam is widely known for its backup application on initially virtual but now physical environments as well. However, the official name of the product is ‘Veeam Backup and Replication’. Indeed, it can also replicate VM’s to another host, onsite or offsite with a WAN in between, giving you protection on VM level.

Since most customers have Veeam deployed, you might consider using the replication feature of Veeam as well. Know that you might want a dedicated replication network for performance reasons when the VM count is rising. Definitely follow the best practices from Veeam there.

In the future, you can even expect the replication interval to shrink when Veeam has implemented CDP (Continuous Data Protection). This could mean going from an 5 minute interval to mere seconds.

Also, this will work fine in a small to medium environment. If you have a larger environment and you also want to dynamically manage this you should look to VAO (Veeam Availability Orchestrator) that will solve your needs.

HPE MSA

Most VSA customers love the synchronous replication of the VSA offering RPO = 0. At that moment MSA was never an option since it does asynchronous replication only…

However, I have a lot of customers that deployed the VSA solution to virtualize their local storage inside their hypervisor servers into a shared storage platform, and where those servers are installed in the same server room. So, no 2 server rooms.

Or when you start asking them a bit further about their data availability requirements they really don’t need an RPO of 0 or 6 nines of guaranteed availability at the end… An RPO of a few minutes is often acceptable for a much lower price…

At that moment HPE MSA is a good and valid option. 2 controllers to protect against planned and unplanned downtime, high performance with the caching and tiering mechanisms (go for the MSA 2052 model which includes all required licenses) and a variety of connectivity (FC, iSCSI and SAS). What else do you need more from a storage solution with a proven track record of more than 10 years?

Application level replication

If you want high availability on application level, you might consider techniques lihe Microsoft Exchange DAG (Database Availability Groups) or Microsoft SQL Replication. Also Oracle deliver same mechanisms to replicate databases…

Won’t go too much into details here since this is independent of the hardware underneath the running software and applications, Google is your friend for more information here: SQL – Exchange – Oracle

Side note: I do not dive deeper in another great storage platform called HPE 3PAR since this is positioned above the typical requirements a VSA customer is looking for, however who knows what the future may bring.

You see? Options enough, depending on those little differences in requirements of your specific environment. And if it is still not clear for your situation, feel free to contact me and we will work out a solution together…

Be social and share

The post HPE StoreVirtual VSA EOL? What are your options… appeared first on BITCON IT consultants.

Last year HPE decided to EOL the StoreVirtual VSA portfolio leaving a big gap in the software defined storage portfolio. IT was filled up partially with the SimpliVity solution although that this is not a true software-only solution, leaving customers behind needing to migrate to other solutions from VMware, Microsoft or other ISV’s.

Until now! StorMagic has been added to the HPE Complete Program, signifying that HPE and its resellers around the world will now offer SvSAN with HPE servers to their customers.

StorMagic SvSAN

For me this choice is quite obvious since the SvSAN solution from StorMagic is quite similar to what the StoreVirtual VSA solution did: transform local storage inside 2 servers into a shared pool of storage accessible via iSCSI. This is all established by deploying a lightweight VM on the hypervisor (what a coincidence they also called it a VSA as well to make our lives easy!) delivering HA for the data stored on it.

Deployment of this VSA is done from a vCenter plugin or with the use of an OVA file. There are also wizards available for the deployment on Windows Hyper-V and CentOS KVM or RHV.

I will publish here soon an article specifically on the deployment and configuration of the SvSAN solution on this website.

Storage layout

The underlying storage as such is not so important (almost anything is supported) although some best practices are out there.

A capacity tier is deployed with SAS or NL-SAS drives in a RAID5 configuration using a decent RAID controller to protect local data per node. Optionally, a caching tier is configured with Mixed-Use SSD’s in RAID1. A recommendation is a 10% ratio on the SSD capacity against the capacity tier. SSD caching is Read-Write so plan RAID protection for the SSD’s.

Memory caching is also supported (Read-only) allowing a reservation of 1GB of RAM per server to enhance the performance even further.

Predictive Storage Caching

Frequently read blocks of data are pulled up into SSD and memory cache. StorMagic calls this Predictive Storage Caching.

This is beneficial, rather than a least frequently read algorithm, due to a typical SvSAN environment having a nightly backup like VEEAM or CommVault run against it.  This would do a full read of the environment and have a big impact on the cache by flushing everything out the bottom.

Besides this, SvSAN also enables read ahead and data pinning into memory.

Networking requirements

SvSAN uses standard TCP networking so it can use standard switches or virtual distributed switches, LACP, Etherchannel etc with various network topologies.

However, the above is common in a configuration with direct connect or back to back cables fully supported in a 2x node architecture.  This can be 10Gb/25Gb connectivity, ethernet or SFP+, back to back that can be shared with live migration/vMotion or Fault Tolerance networking, while still having slower switched connections out the front of the systems.

A typical configuration will be 2 x DL380 servers with a 10Gb dual-port LOM card for direct-attach storage network connectivity (without switches) and the 4x1Gb network ports will be used for management and VM networking.

Quorum Witness tie-breaker

Similar to the FOM aka FailOver Manager function with StoreVirtual is StorMagic using as well a tie-breaker to avoid split-brain situations to avoid data inconsistency. This is done by deploying a SvSAN Witness function within the network. This can be done through a Windows service, a linux deamon or using a packaged VM. It can even run on a Raspberry Pi, and a single witness can serve 1000s of mirrors. Network latency of 3000 ms is tolerated meaning it could be on the other side of the workd. Bandwidth requirements are also lightweight, barely 9kbps… Impressive!

Licensing

Licensing is quite simple: what is the required capacity and you want caching for ultimate performance or not?

SvSAN is available in two editions – “Standard” and “Advanced”.

• Standard Edition: Synchronous mirroring of your local data across 2 servers accessible via and iSCSI target.
• Advanced Edition: Adds key performance features collectively known as Predictive Storage Caching.

SvSAN is a capacity license, based on the amount of usable disk capacity presented to SvSAN, NOT the raw disk capacity of the servers. Licenses are available in four capacity levels:

• 2TB
• 6TB
• 12TB
• Unlimited

The licensed usable capacity can be increased to a higher capacity level at any time. However, it is not possible to reduce capacity to a lower level. SvSAN is sold as a perpetual license in pairs of license keys. Each Virtual Storage Appliance (VSA) requires a license.

Differences

Although that there are quite some similarities with the StoreVirtual VSA solution there are also some differences. The StoreVirtual solution was a scale-out solution, starting with 2 nodes and being able to grow the capacity and performance by adding nodes in the cluster.

The StorMagic solution is always a 2-node solution enabling mirroring of a volumes across these 2 nodes. A 3-node setup is the maximum configuration allowing the synchronous mirroring of multiple targets across these 3 nodes. That’s it.

I don’t see this as an issue however since there is an unlimited capacity license available, the maximum capacity will be defined by the maximum number of disks inside a server. With servers allowing these days +30 disks I don’t see directly a challenge here.

Further, not that I was the biggest fan of the CMC Centralized Management Console but something that I miss is a centralized console. Not really a complaint but now I need to use 3 interfaces depending of what I want to do. Daily operations are done through the very basic vCenter plug-in like deploying VSA’s (works like a charm) and creating volumes (could it be easier) and connect them to hosts. The techies like me however will need to connect to the web interface of each VSA separately to get more details about the configuration of the storage solution.

More information

Trial software is available via the StorMagic website: https://stormagic.com/trial

If you want more information or a demo, feel free to contact us!

I have a new challenge: today I started counting to get as well at least 12 years of 100% availability with my StorMagic volume just as I did with my StoreVirtual VSA volume, which is still active and counting!

Be social and share!

The post HPE introduces StorMagic SvSAN as StoreVirtual VSA replacement appeared first on BITCON IT consultants.