Storage Field Day 10 – Wrap-up and Link-o-rama

Disclaimer: I recently attended Storage Field Day 10.  My flights, accommodation and other expenses were paid for by Tech Field Day. There is no requirement for me to blog about any of the content presented and I am not compensated in any way for my time at the event.  Some materials presented were discussed under NDA and don’t form part of my blog posts, but could influence future discussions.

SFD-Logo2-150x150

This is a quick post to say thanks once again to Stephen, Tom, Megan and the presenters at Storage Field Day 10. I had an enjoyable and educational time. For easy reference, here’s a list of the posts I did covering the event (they may not match the order of the presentations).

Storage Field Day – I’ll Be At SFD10

Storage Field Day 10 – Day 0

Storage Field Day 10 – (Fairly) Full Disclosure

Kaminario are doing some stuff we’ve seen before, but that’s okay

Pure Storage really aren’t a one-trick pony

Tintri Keep Doing What They Do, And Well

Nimble Storage are Relentless in Their Pursuit of Support Excellence

Cloudian Does Object Smart and at Scale

Exablox Isn’t Just Pretty Hardware

It’s Hedvig, not Hedwig

The Cool Thing About Datera Is Intent

Data Virtualisation is More Than Just Migration for Primary Data

 

Also, here’s a number of links to posts by my fellow delegates (and Tom!). They’re all really quite smart, and you should check out their stuff, particularly if you haven’t before. I’ll try keep this updated as more posts are published. But if it gets stale, the SFD10 landing page has updated links.

 

Chris M Evans (@ChrisMEvans)

Storage Field Day 10 Preview: Hedvig

Storage Field Day 10 Preview: Primary Data

Storage Field Day 10 Preview: Exablox

Storage Field Day 10 Preview: Nimble Storage

Storage Field Day 10 Preview: Datera

Storage Field Day 10 Preview: Tintri

Storage Field Day 10 Preview: Pure Storage

Storage Field Day 10 Preview: Kaminario

Storage Field Day 10 Preview: Cloudian

Object Storage: Validating S3 Compatibility

 

Ray Lucchesi (@RayLucchesi)

Surprises in flash storage IO distributions from 1 month of Nimble Storage customer base

Has triple parity Raid time come?

Pure Storage FlashBlade well positioned for next generation storage

Exablox, bring your own disk storage

Hedvig storage system, Docker support & data protection that spans data centers

 

Jon Klaus (@JonKlaus)

I will be flying out to Storage Field Day 10!

Ready for Storage Field Day 10!

Simplicity with Kaminario Healthshield & QoS

Breaking down storage silos with Primary Data DataSphere

Cloudian Hyperstore: manage more PBs with less FTE

FlashBlade: custom hardware still makes sense

Squashing assumptions with Data Science

Bringing hyperscale operations to the masses with Datera

Making life a whole lot easier with Tintri VM-aware storage

 

Enrico Signoretti (@ESignoretti)

VM-aware storage, is it still a thing?

Scale-out, flash, files and objects. How cool is Pure’s FlashBlade?

 

Josh De Jong (@EuroBrew)

 

Max Mortillaro (@DarkkAvenger)

Follow us live at Storage Field Day 10

Primary Data: a true Software-defined Storage platform?

If you’re going to SFD10 be sure to wear microdrives in your hair

Hedvig Deep Dive – Is software-defined the future of storage?

Pure Storage’s FlashBlade – Against The Grain

Pure Storage Flashblade is now available!

 

Gabe Maentz (@GMaentz)

Heading to Tech Field Day

 

Arjan Timmerman (@ArjanTim)

We’re almost live…

Datera: Elastic Data Fabric

 

Francesco Bonetti (@FBonez)

EXABLOX – A different and smart approach to NAS for SMB

 

Marco Broeken (@MBroeken)

 

Rick Schlander (@VMRick)

Storage Field Day 10 Next Week

Hedvig Overview

 

Tom Hollingsworth (@networkingnerd)

Flash Needs a Highway

 

Finally, thanks again to Stephen, Tom, Megan (and Claire in absentia). It was an educational and enjoyable few days and I really valued the opportunity I was given to attend.

SFD10_GroupPhoto

Data Virtualisation is More Than Just Migration for Primary Data

Disclaimer: I recently attended Storage Field Day 10.  My flights, accommodation and other expenses were paid for by Tech Field Day. There is no requirement for me to blog about any of the content presented and I am not compensated in any way for my time at the event.  Some materials presented were discussed under NDA and don’t form part of my blog posts, but could influence future discussions.

logo-Primary_Data-400px

Before I get started, you can find a link to my raw notes on Primary Data’s presentation here. You can also see videos of the presentation here. I’ve seen Primary Data present at SFD7 and SFD8, and I’ve typically been impressed with their approach to Software-Defined Storage (SDS) and data virtualisation generally. And I’m also quite a fan of David Flynn‘s whiteboarding chops.

SFD10_Pd_DavidFlynn

 

Data Virtualisation is More Than Just Migration

Primary Data spent  some time during their presentation at SFD10 talking about Data Migration vs Data Mobility.

SFD10_Pd_DataMigrationvsMobility

[image courtesy of Primary Data]

Data migration can be a real pain to manage. It’s quite often a manual process and is invariably tied to the capabilities of the underlying storage platform hosting the data. The cool thing about Primary Data’s solution is that it offers dynamic data mobility, aligning “data’s needs (objectives) with storage capabilities (service levels) through automated mobility, arbitrated by economic value and reported as compliance”. Sounds like a mouthful, but it’s a nice way of defining pretty much what everyone’s been trying to achieve with storage virtualisation solutions for the last decade or longer.

What I like about this approach is that it’s a data-centric, rather than employing a storage platform focused approach. Primary Data supports “anything that can be presented to Linux as a block device”, so the options to deploy this stuff are fairly broad. Once you’ve presented your data to DSX, there’s some smart service level objectives (SLOs) that can be applied to the data. These can be broken down into the categories of protection, performance, and price/penalty:

Protection

  • Durability
  • Availability
  • Recoverability – Security
  • Priority
  • Sovereignty

Performance

  • IOPS / Bandwidth / Latency – Read / Write
  • Sustained / Burst

Price / Penalty

  • Per File
  • Per Byte
  • Per Operation

Access Control can also be applied to your data. With Primary Data, “[e]very storage container is a landlord with floorspace to lease and utilities available (capacity and performance)”.

 

Further Reading and Final Thoughts

I like the approach to data virtualisation that Primary Data have taken. There are a number of tools on the market that claim to fully virtualise storage and offer mobility across platforms. Some of them do it well, and some focus more on the benefits provided around ease of migration from one platform to another.

That said, there’s certainly some disagreement in the market place on whether Primary Data could be considered a fully-fledged SDS solution. Be that as it may, I really like the focus on data, rather than silos of storage. I’m also a big fan of applying SLOs to data, particularly when it can be automated to improve the overall performance of the solution and make the data more accessible and, ultimately, more valuable.

Primary Data has a bunch of use cases that extend beyond data mobility as well, including deployment options ranging from Hyperconverged, software-defined NAS and clustering across existing storage platforms. Primary Data want to “do for storage what VMware did for compute”. I think the approach they’ve taken has certainly gotten them on the right track, and the platform has matured greatly in the last few years.

If you’re after some alternative (and better thought out) posts on Primary Data, you can read Jon‘s post here. Max also did a good write-up here, while Chris M.Evans did a nice preview post on Primary Data that you can find here.

The Cool Thing About Datera Is Intent

Disclaimer: I recently attended Storage Field Day 10.  My flights, accommodation and other expenses were paid for by Tech Field Day. There is no requirement for me to blog about any of the content presented and I am not compensated in any way for my time at the event.  Some materials presented were discussed under NDA and don’t form part of my blog posts, but could influence future discussions.

Datera-250x33

Before I get started, you can find a link to my raw notes on Datera‘s presentation here. You can also see videos of their presentation here.

 

What’s a Datera?

Datera’s Elastic Data Fabric is “software defined storage appliance that takes over the hardware”. It’s currently available in two flavours:

  • Software available with qualified hardware (this is prescriptive, and currently based on a SuperMicro platform); and
  • Can be licensed as software-only as well with 2 SKUs available in 50TB or 100TB chunks.

 

What Can I Do With a Datera?

SFD10_Datera

[image courtesy of Datera]

There are a couple of features that make Datera pretty cool, including:

  • Intent defined – you can use templates to enable intelligent placement of application data;
  • Economic flexibility – heterogeneous nodes can be deployed in the same cluster (capacity, performance, media type);
  • Works with an API first or Dev/Ops model – treating your infrastructure as code, programmable/composable;
  • Multi-tenant capability – this includes network isolation and QoS features;
  • Infrastructure awareness – auto-forming, optimal allocation of infrastructure resources.

 

What Do You Mean “Intent”?

According to Datera, Application Intent is “[a] way of describing what your application wants and then letting the system allocate the data”. You can define the following capabilities with an application template:

  • Policies for management (e.g. QoS) – data redundancy, data protection, data placement;
  • Storage template – defines how many volumes you want and the size you want; and
  • Pools of resources that will be consumed.

I think this is a great approach, and really provides the infrastructure operator with a fantastic level of granularity when it comes to deploying their applications.

Datera don’t use RAID, currently using 1->5 replication (synchronous) within the cluster to protect data. Snapshots are copy on write (at an application intent level).

Further Reading and Final Thoughts

I know I’ve barely scratched the surface of some of the capabilities of the Datera platform. I am super enthusiastic about the concept of Application Intent, particularly as it relates to scale-out, software-defined storage platforms. I think we spend a lot of time talking about how fast product X can go, and why technology Y is the best at emitting long beeps or performing firmware downgrades. We tend to forget about why we’re buying product X or deploying technology Y. It’s to run the business, isn’t it? Whether it’s teaching children or saving lives or printing pamphlets, the “business” is the reason we need the applications, and thus the reason we need the infrastructure to power those applications. So it’s nice to see vendors such as Datera (and others) working hard to build application-awareness as a core capability of their architecture. When I spoke to Datera, they had four customers announced, with more than 10 “not announced”. They’re obviously keen to get traction, and as their product improves and more people get to know about them, I’ve no doubt that this number will increase dramatically.

While I haven’t had stick-time with the product, and thus can’t talk to the performance or otherwise, I can certainly vouch for the validity of the approach from an architectural perspective. If you’re looking to read up on software-defined storage, I wouldn’t hesitate to recommend Enrico‘s recent post on the topic. Chris M. Evans also did a great write-up on Datera as part of his extensive series of SFD10 preview posts – you can check it out here. Finally, if you ever need to get my attention in presentations, the phrase “no more data migration orgies” seems to be a sure-fire way of getting me to listen.

It’s Hedvig, not Hedwig

Disclaimer: I recently attended Storage Field Day 10.  My flights, accommodation and other expenses were paid for by Tech Field Day. There is no requirement for me to blog about any of the content presented and I am not compensated in any way for my time at the event.  Some materials presented were discussed under NDA and don’t form part of my blog posts, but could influence future discussions.

hedvig_logo_260x77-200x59

Before I get started, you can find a link to my raw notes on Hedvig‘s presentation here. You can also see videos of the presentation here.

 

It’s Hedvig, not Hedwig

I’m not trying to be a smart arse. But when you have a daughter who’s crazy about Harry Potter, it’s hard not to think about Hedwig when seeing the Hedvig brand name. I’m sure in time I’ll learn not to do this.

If you’re unfamiliar with Hedvig, it’s software-defined storage. The Hedvig Distributed Storage Platform is made up of standard servers and the Hedvig software.

Some of the key elements of the Hedvig solution are as follows:

  • Software is completely decoupled from commodity hardware;
  • Application-specific storage policies; and
  • Automated and API-driven.

 

Capabilities

Hedvig took us through their 7 core capabilities, which were described as follows:

  • Seamless scaling with x86 or ARM (haven’t seen an ARM-64 deployment yet);
  • Hyperconverged and hyperscale architectures (can mix and match in the same cluster);
  • Support for any hypervisor, container or OS (Xen, KVM, HyperV, ESX, containers, OpenStack, bare-metal Windows or Linux);
  • Block (iSCSI), file (NFS) and object (S3, SWIFT) protocols in one platform;
  • Enterprise features: dedupe, compression, tiering, caching, snaps/clones;
  • Granular feature provisioning per virtual disk; and
  • Multi-DC and cloud replication.

 

Components

SFD10_Hedvig_Components

The Hedvig solution is comprised of the following key components:

  • Hedvig Storage Proxy – presents the block and file storage; runs as VM, container, or bare metal;
  • Hedvig Storage Service – forms an elastic cluster using commodity servers and/or cloud infrastructure; and
  • RESTful APIs – provides object access via S3 or Swift, instruments control and data plane

 

How Does It Work?

This is oversimplifying things, but here’s roughly how it works:

  • Create and present virtual disks to the application tier;
  • Hedvig Storage Proxy captures and directs I/O to storage cluster;
  • Hedvig Storage Service distributes and replicates data across nodes;
  • The cluster caches and balances across nodes and racks; and
  • The cluster replicates for DR across DCs and/or clouds.

 

Use Cases?

So where would you use Hedvig? According to Hedvig, they’re seeing uptake in a number of both “traditional” and “new” areas:

Traditional

  • Server virtualisation
  • Backup and BC/DR
  • VDI

New workloads

  • Production clouds
  • Test/Dev
  • Big data/IoT

 

Further Reading and Final Thoughts

Before I wrap up, a quick shout-out to Chris Kranz for his use of Hedvig flavoured magnetic props during his whiteboard session – it was great. Here’s a shonky photo of Chris.

SFD10_Hedvig

Avinash Lakshman is a super smart dude with a tonne of experience in doing cloud and storage things at great scale. He doesn’t believe that traditional storage has a future. When you watch the video of the Hedvig presentation at SFD10 you get a real feel for where the company’s coming from. The hyper-functional API access versus the GUI that looks a little rough around the edges certainly gives away the heritage of this product. That said, I think Avinash and Hedvig are onto a good thing here. The “traditional” storage architectures are indeed dying, as much as we might enjoy the relative simplicity of selling someone a dual-controller, midrange, block array with limited scalability.

As with many of these solutions I feel like we’re on the cusp of seeing something really cool being developed right in front of us. For some us, the use cases won’t strike a chord, and the need for this level of scalability may not be there. But if you’re all in on SDS, Hedvig certainly has some compelling pieces of the puzzle that I think are worthy of further investigation.

The Hedvig website contains a wealth of information. You should also check out Chris M. Evans‘s SFD10 preview post on Hedvig here, while Rick Schlander did a great overview post that I recommend reading. Max did a really good deep dive post, along with a higher level view that you can see here.

 

Exablox Isn’t Just Pretty Hardware

Disclaimer: I recently attended Storage Field Day 10.  My flights, accommodation and other expenses were paid for by Tech Field Day. There is no requirement for me to blog about any of the content presented and I am not compensated in any way for my time at the event.  Some materials presented were discussed under NDA and don’t form part of my blog posts, but could influence future discussions.

exablox-logo-black

Before I get started, you can find a link to my raw notes on Exablox‘s presentation here. You can also see videos of the presentation here.  You can find a preview post from Chris M. Evans here.

 

It’s Not Just the Hardware

I waxed lyrical about the Exablox hardware platform after seeing it at Storage Field Day 7. But while the OneBlox hardware is indeed pretty cool (you can see the specifications here), the cloud-based monitoring platform, OneSystem, is really the interesting bit.

According to Exablox, the “OneSystem application is used to combine OneBlox appliances into Rings as well as configuring shares, user access, and remote replication”. It’s the mechanism used for configuration, as well as monitoring, alerting and reporting.

OneSystem is built on a cloud-based, multi-tenant architecture. There’s nothing to install for organisations, VARs, and MSPs. Although if you feel a bit special about how your data is treated, there is an optional, private OneSystem deployment available for on-premises management. Exablox pride themselves on the “world-class” support they provide to customers, with a customer-first culture being one of the dominant themes when talking to them about support capability. Some of the other benefits of the OneSystem approach is:

  • The ability to globally manage OneBlox anywhere; and
  • Deliver seamless OneBlox software upgrades.

Exablox also provide 24×7 proactive monitoring, providing insight into, amongst other things:

  • Storage utilisation and analysis;
  • Storage health and alerts; and
  • OneBlox drive health.

The cool thing about this platform is that it offers the ability to configure custom storage policies and simple scaling for individual applications. In this manner you can configure the following data services on a “per application” basis:

  • Variable or fixed-length deduplication;
  • Compression on/off;
  • Continuous data protection on/off and retention; and
  • Remote replication on/off.

 

I Want My Data Everywhere

While the OneBlox ring is currently limited to 7 systems per cluster, you can have two or more (up to 10) clusters operating in a mesh for replication. You can then conceivably have a whole bunch of different data protection schemes in place depending on what you need to protect and where you need it protected. The great thing is that, with the latest version of OneSystem, you can have a one-to-many replication relationship between directories as well. This kind of flexibility is really neat in my opinion. Note that replication is asynchronous.

SFD10_Exablox_Mutli-siteReplication

 

Further Reading and Final Thoughts

If you’ve read any of my recent posts on the likes of Pure, Nimble and Tintri, it would feel like everyone and their dog is into cloud-based monitoring and analytics systems for storage platforms. This is in no way a bad thing, and something that I’m glad we’re seeing become a prevalent feature with these “modern” storage architectures. We store a whole bunch of data on these things. And sometimes it’s even data that is vital to the success of the various business endeavours we undertake on a daily basis. So it’s great to see vendors are taking this requirement seriously. It also helps somewhat that people are a little more comfortable with the concept of keeping information in “the cloud”. This certainly helps the vendors control the end user experience form a support viewpoint, rather than relyin on arcane systems deployed across multiple VMs that invariably fail at the time you need to dig into the data to find out what’s really going on in the environment.

Exablox have come up with a fairly unique approach to scale-out NAS, and I’m keen to see where they take it from here. Features such as remote replication and the continuing maturity of the OneSystem platform make me think that they’re gearing up to push things a little beyond the BYO drives SMB space. I’ll be interested to see just how that plays out.

Ray Lucchesi did a thorough write-up on Exablox that you can read here, while Francesco Bonetti did a great write-up here. Exablox has also published a technical overview of OneBlox and OneSystem that is worth checking out.

 

Cloudian Does Object Smart and at Scale

Disclaimer: I recently attended Storage Field Day 10.  My flights, accommodation and other expenses were paid for by Tech Field Day. There is no requirement for me to blog about any of the content presented and I am not compensated in any way for my time at the event.  Some materials presented were discussed under NDA and don’t form part of my blog posts, but could influence future discussions.

cloudian-logo

Before I get started, you can find a link to my raw notes on Cloudian‘s presentation here. You can also see videos of the presentation here.

I’m quite keen on Cloudian’s story, having seen them in action at Storage Field Day 7. I also got to have a beer with Michael Tso at the SFD10 mixer and talk all things Australian.

SFD10_Cloudian_MichaelTso_Cropped

 

Smart and at Scale

Cloudian took us through some of their driving design principles, and I thought it was worth covering these off again. You’ll notice the word “scale” gets used a lot, and this has been a particularly important capability for Cloudian. They did a blog post on it too.

One of the key features of the HyperStore solution is that it needed to support what Cloudian term “Smart Operations at Scale”. This requires the tech to:

  • Be simple and intuitive;
  • Be fully automated from an operations perspective (e.g. add/remove drives/nodes, upgrades);
  • Provide visual storage analytics to automatically see hot spots; and
  • Offer self service consumption (via a policy based approach).

Cloudian have also worked hard to ensure they can provide “Extreme Durability at Scale”, with the HyperStore solution offering the ability to:

  • Be always repaired, always verified;
  • Offer automated failure avoidance (through the use of Dynamic Object Routing); and
  • Be “enterprise” grade.

One of the keys to being able deliver a scaleable solution has been the ability to provide the end user with “Smart support at Scale”, primarily through the use of:

  • Proactive (not reactive) support;
  • Continuous monitoring; and
  • Global analytics.

The analytics piece is a big part of the Cloudian puzzle, and something they’ve been working hard on recently. With their visual analytics you can analyse your data across globe and plan for future based on your demand. Cloudian not only performs analytics at scale, but also designed to facilitate operations at scale, with:

  • One screen for hundreds of nodes (in a kind of “beehive” layout);
  • Instant view of a node’s health;
  • The ability to add nodes with one click; and
  • The ability to dynamically rebalance the cluster.

When it comes to software defined storage platforms, the simple things matter, particularly as it relates to your interactions with the hardware platform. To that end, with HyperStore you’ve got the ability to do some basic stuff, like:

  • Identifying node types;
  • Blinking suspect servers; and
  • Blinking suspect drives.

When you’re running a metric s**t-tonne of these devices in a very big data centre, this kind of capability is really important, especially when it comes to maintenance. As is the ability to perform rolling upgrades of the platform with no downtime and in an automated fashion. When it comes to rebuilds, Cloudian provides insight into both data rebuild information and cluster rebalance information – both handy things to know when something’s gone sideways.

The Cloudian platform also does “Smart Disk Balancing”. If there’s a disk imbalance it will change the tokens pointing from “highly used disk to low used disk”. If there’s a disk failure, new data automatically routes to newly assigned resources. Makes sense, and nice to see they’ve thought it through.

 

Further Reading and Conclusion

Cloudian make quite a big deal of their S3 compatibility. They even give me a sticker that says it’s guaranteed. It looks a lot like this:

Badge_S3YourDataCenter_transparent2

Chris Evans also did a series of posts on S3 and Cloudian that you can read here, here and here. He also did a great preview post prior to SFD10 which is also worth a look. He’s a good lad, he is. Particularly when I need to point you, my loyal reader, to well written articles on topics I’m a little sketchy on.

S3 compatibility is a big thing for a lot of people looking at deploying object storage, primarily because AWS are leaps and bounds ahead of the pack in terms of object storage functionality, deployed instances, and general mindshare. Cloudian haven’t just hitched their wagon to S3 compatibility though. In my opinion they’ve improved on the S3 experience through clever design and a solid approach to some fundamental issues that arise when you’re deploying a whole bunch of devices in data centres that don’t often have staff members present.

Nimble Storage are Relentless in their Pursuit of Support Excellence

Disclaimer: I recently attended Storage Field Day 10.  My flights, accommodation and other expenses were paid for by Tech Field Day. There is no requirement for me to blog about any of the content presented and I am not compensated in any way for my time at the event.  Some materials presented were discussed under NDA and don’t form part of my blog posts, but could influence future discussions.

400px-Nimble_logo

Before I get cracking, you can find a link to my raw notes on Nimble Storage‘s presentation here. You can also see videos of the presentation here.

I’ve written about Nimble recently. I went to their Predictive Flash Platform launch in San Francisco earlier this year. You can read about that here (disclosure is here). I’ve also talked about InfoSight with some level of enthusiasm. I think this all ties in nicely with my thoughts on their SFD10 presentation.

Before I get into that though, kudos to Tom McKnight (VP Hardware Engineering) for his demo on component resilience (pulling 6 drives, PSU and causing a controller failure). Demos are tough at the best of times, and it’s always nice to see people with the confidence to stand behind their product and run it through its paces in front of a “live studio audience”.

SFD10_Nimble_TomMcKnight

 

Tier 3 Before You Know It

Rod Bagg (VP Analytics and Customer Support) provided an overview of InfoSight. He spoke a lot about what he called the “app-data gap”, with the causes of problems in the environment being:

  • Storage related;
  • Configuration issues;
  • Non-storage best practices;
  • Interoperability issues; and
  • Host, compute, VM, etc.

But closing the app-data gap with tech (in this case, SSDs) oftentimes is not enough. You need predictive analytics. Every week InfoSight analyses more than a trillion data points. And it’s pretty good in helping you make your infrastructure transparent. According to Rod, it:

  • Proactively informs and guides without alarm fatigue;
  • Predicts future needs and simplifies planning; and
  • Delivers transformed support experience from Level 3 experts.

Nimble say that 9 out of 10 issues are detected before you know about them. “If we know about an issue, it shouldn’t happen to you”. Rod also spoke at some length about the traditional Level 3 Support model vs. Nimble’s approach. He said that you could “pick up the phone, dial 1-877-364-6253, and get Level 3 Support”, with the average hold time being <1 minute. This isn’t your standard vendor support experience, and Nimble were very keen to remind us of that.

SFD10_Nimble_TraditionalSupport

 

Further Reading and Conclusion

I’ve said before that I think InfoSight is a really cool tool. It’s not just about Nimble’s support model, but the value of the data they collect and what they’re doing with that data to solve support issues in a proactive fashion. It also provides insight (!) into what customers are doing out in the real world with their arrays. Ray Lucchesi had a nice write-up on IO distribution here that is well worth a read. Chris M. Evans also did a handy preview post on Nimble that you can find here.

Whenever people have asked me in the past what they should be looking for in a storage array, I’ve been reluctant to recommend vendors based purely on performance specifications or the pretty bezel. When I was working in operations, the key success criterion for me was the vendor’s ability to follow up on issues with reliable, prompt support. Nothing works perfectly, despite what vendors tell you. Having the ability to fix things in a timely fashion, through solid logistics, good staff and a really solid analytics platform, provides vendors like Nimble with an advantage over their competitors. Indeed, a few other vendors, including Pure and Kaminario, have seen the value in this approach and are taking similar approached with their support models. It will be really interesting to see how the platform evolves over time and how Nimble’s relentless pursuit of support excellence scales as the company grows bigger.

Tintri Keep Doing What They Do, And Well

Disclaimer: I recently attended Storage Field Day 10.  My flights, accommodation and other expenses were paid for by Tech Field Day. There is no requirement for me to blog about any of the content presented and I am not compensated in any way for my time at the event.  Some materials presented were discussed under NDA and don’t form part of my blog posts, but could influence future discussions.

Tintri_Logo_Horizontal_1024

Before I get into it, you can find a link to my notes on Tintri‘s presentation here. You can also see videos of the presentation here.

I’ve written about Tintri recently. As recently, in fact, as a week before I saw them at SFD10. You can check out my article on their most recent product announcements here.

 

VAS but not AAS (and that’s alright)

Tintri talk a lot about VM-aware Storage (or VAS as they put it). There’s something about the acronym that makes me cringe, but the sentiment is admirable. They put it all over their marketing stuff. They’re committed to the acronym, whether I like it or not. But what exactly is VM-aware Storage? According to Tintri, it provides:

  • VM-level QoS;
  • VM-level analytics;
  • VM data management;
  • VM-level automation with PowerShell and REST; and
  • Supported across multiple hypervisors (Support VMware, Hyper-V, OpenStack, RedHat).

Justin Lauer, Global Evangelist with Tintri, took us through a demo of VAS and the QoS capabilities built in to the Tintri platform.

SFD10_Tintri_Justin

I particularly liked the fact that I can get a view of end to end latency (host / network / storage (contention and flash) / throttle latency). In my opinion this is something that people have struggled with for some time, and it looks like Tintri have a really good story to tell here. I also liked the look of the “Capacity gas gauge” (petrol for Antipodeans), providing an insight into when you’ll run out of either performance, capacity, or both.

So what’s AAS then? Well, in my mind at least, this is the ability to delve into application-level performance and monitoring, rather than just VM-level. And I don’t think Tintri are doing that just yet. Which, to my way of thinking, isn’t a problem, as I think a bunch of other vendors are struggling to really do this in a cogent fashion either. But I want to know what my key web server tier is doing, for example, and I don’t want to assume that it still lives on the datastore that I tagged for it when I first deployed it. I’m not sure that I get this with VAS, but I still think it’s a long way ahead of where we were a few years ago, getting stats out of volumes and not a lot else.

 

Further Reading and Conclusion

In the olden days (a good fifteen years ago) I used to struggle to get multiple Oracle instances to play nicely on the same NT4 host. But I didn’t have a large number of physical hosts to play with, and I had limited options when I wanted to share resources across applications. Virtualisation to slice up physical resources in a more concise fashion, And as a result of this it’s made it simple for us to justify running one application per VM. In this way we can still get insights into our applications from understanding what our VMs are doing. This is no minor thing when you’re looking after storage in the enterprise – it’s a challenge at the best of times. Tintri has embraced the concept of intelligent analytics in their arrays in the same way that Nimble and Pure have started really making use of the thousands of data points that they collect every minute.

But what if you’re not running virtualised workloads? Well, you’re not going to get as much from this. But you’ve probably got a whole lot of different requirements you’re working to as well. Tintri is really built from the ground up to deliver insight into virtualised workloads that has been otherwise unavailable. I’m hoping to see them take it to the next level with application-centric monitoring.

Finally, Enrico did some more thorough analysis here that’s worth your time. And Chris’s SFD10 preview post on Tintri is worth a gander as well.

 

Pure Storage really aren’t a one-trick pony

Disclaimer: I recently attended Storage Field Day 10.  My flights, accommodation and other expenses were paid for by Tech Field Day. There is no requirement for me to blog about any of the content presented and I am not compensated in any way for my time at the event.  Some materials presented were discussed under NDA and don’t form part of my blog posts, but could influence future discussions.

PureStorage_logo

Before I get started, you can find a link to the Pure Storage presentation at Storage Field Day 10 here. You can also see the videos of the presentation here.

In talking with people about Pure Storage, some of the feedback I received was that they were a “one-trick pony“. By that I mean that people thought all they did was offer an all-flash array and nothing more. I think there’s always been a lot more to Pure than just the array. To wit, their approach to hardware maintenance and lifecycles, via the Evergreen Storage program, as well as their implementation of Pure1 has had me thinking for a while that they’re not your father’s AFA vendor.

 

FlashBlade

I wrote about FlashBlade when it was first announced and I was cautiously optimistic that they were onto something kind of cool. As it happened, Pure spent a lot of time at SFD10 giving us a run-through on what some of the thinking around the design of the FlashBlade was, and it solidified some of my ideas around the product.

Here’s a happy snap of Brian Gold (@briantgold) taking us through the hardware in the FlashBlade.

Pure_SFD10

Some of the challenges Pure aimed to address when designing the FlashBlade was the need for scale in terms of:

  • Capacity – from terabytes to petabytes;
  • Concurrency – from a few users to thousands; and
  • Access patterns – from small files and metadata to large, streaming workloads.

They also wanted to do this without drowning the users or administrators in complexity.

One of the key approaches to this problem was to adopt a modular architecture through the use of the blade chassis. While we talk a lot about the flash in Pure’s FlashBlade, the network architecture shouldn’t be underestimated. A key component of Pure’s “software-defined networking” is hardware (no, the irony is not lost on me), with two Broadcom Trident-II ethernet switch ASICs collapsing three networks (Front End, Back End and Control) into one high performance fabric providing 8 40Gbs QSFP connections into customer Top of Rack switches. This provides Pure with the use of a high performance, integrated fabric connected to scalable server nodes. While some of the specifications at the time of announcement were limited to the chassis, you’ll start to see these numbers increase as the SDN component is improved over time.

Brian was keen to see us thinking about the FlashBlade hardware design in the following terms:

  • An integrated blade chassis provides density and simplicity;
  • All-flash storage unlocks the parallelism inside an SSD; and
  • An NVRAM engine built for distributed transaction processing.

Rob Lee then went on to talk about the software side of the equation, with the key takeaways from the software side of things being Pure’s desire to:

  • Achieve scalability though parallelism at all layers;
  • Create parallelism through deep partitioning and distribution; and
  • Minimise the cost of distributed coordination.

 

Further Reading and Conclusion

Chris Evans did a nice article on Pure prior to SFD10. Chris Mellor did a decent write-up (something he’s prone to) at the time of release, and Enrico put together some interesting insights as well. Pure are certainly bucking the trend of commodity hardware by using their own stuff. They’re doing scale out differently as well, which is something some pundits aren’t entirely pleased about. All that said, I think the next 12 months will be critical to the success of the scale-out file and object play. Pure’s ability to execute on a technically compelling roadmap, as well as grabbing the interest of customers in rich media, analytics and technical computing will be the ultimate measure of what looks to be a well thought out product architecture. If nothing else, they’ve come up with a chassis that does this …

Kaminario are doing some stuff we’ve seen before, but that’s okay

Disclaimer: I recently attended Storage Field Day 10.  My flights, accommodation and other expenses were paid for by Tech Field Day. There is no requirement for me to blog about any of the content presented and I am not compensated in any way for my time at the event.  Some materials presented were discussed under NDA and don’t form part of my blog posts, but could influence future discussions.

kaminario

For each of the xFD-related posts I do I like to include a few things, namely a link to my presentation notes and a link to the videos of the presentation.

I wrote a brief summary of my last encounter with Kaminario at Storage Field Day 7 – you can check it out here. In this post I’d like to look more into what they’re doing from an analytics and array intelligence perspective. But before I do that I’d like to tip my hat to Shachar Fienblit. Fienblit did a great presentation on what’s coming in storage technology (from Kaminario’s perspective on the industry) and I urge you to check out the video.

SFD10_Kaminario_Fienblit

So it’s probably not really fair to say that Kaminario “are doing some stuff we’ve seen before, but that’s okay”. A better description might be that I think there are thematic commonalities between Kaminario’s approach and Nimble, Tintri and Pure’s various approaches to array analytics and understanding workload.

 

HealthShield

Kaminario have been developing a vision of what they call “DC-aware” storage, with their key imperative being to “simplify and optimise storage deployment and the integration between layers of the IT stack without compromising on cost”. In this fashion they say they differentiate themselves from the current hyperconverged infrastructure paradigm. So what are Kaminario doing?

Firstly, HealthShield is helping them to make some interesting design decisions for the K2 product. What’s HealthShield? It’s “a cloud-based, call-home and analytics engine that delivers proactive monitoring and troubleshooting. Tightly integrated with Kaminario’s world-class support, HealthShield complements high-availability features by ensuring hardware failures never impact availability”. As part of Kaminario’s strategy around understanding the infrastructure better, they’re looking to extend their HealthShield Analytics infrastructure to ensure that storage is both optimised and aware of the full IT deployment.

To this end, with HealthShield they’ve created a Software-as-a-Service offering that helps to “consolidate, analyse, optimise, and automate storage configurations”. While Kaminario’s technical area of expertise is with their storage arrays, they’re very keen to extend this capability beyond the storage array and into the broader application and infrastructure stack. This can only be a good thing, as I think array vendors historically have done something of a shoddy job at understanding what the environment is doing beyond simple IO measurements. The cool thing about making this a SaaS offering is that, as Kaminario say, you can do “Analytics in the cloud so your controllers can work on IO”. Which is a good point too, as we’ve all witnessed the problems that can occur when your storage controllers are working hard on processing a bunch of data points rather than dishing up the required throughput for your key applications. Both Pure and Nimble do this with Pure1 and InfoSight, and I think it’s a very sensible approach to managing your understanding of your infrastructure’s performance. Finally, HealthShield collects thousands of data points, and partners once the customer gives permission.

 

So What Can I Do With This Data?

Shai Maskit, Director of Technical Marketing (pictured below) did a great demo on Kaminario’s Quality of Service (QoS) capabilities that I think added a level of clarity to the HealthShield story. While HealthShield can be a great help to Kaminario in tuning their arrays, it also provides insight into the right settings to use when applying QoS policies.

SFD10_Kaminario_Demo

But what’s the problem with QoS that Kaminario are trying to solve? In Kaminario’s opinion, existing QoS solutions are complicated to administer and don’t integrate into the broader set of application delivery operations. Kaminario have set out to do a few things.

Firstly, they want to simplify storage QoS. They do this by abstracting QoS based on customer-defined policies. In this scenario, the customer also defines preferences, not just how to implement QoS in the environment. The key benefit of this approach is that you can then integrate QoS with the application, allowing you to set QoS polices for specific workloads (e.g. OLTP vs OLAP), while closing the gap between the database and its storage platform.

Another key benefit is the availability of performance data, with analytics being made available to detect changing performance patterns and automatically adapt. This also provides insight into workload migration to the K2 environment based on application performance. This can be extremely handy when you don’t want to run everything on your all flash array.

 

Conclusion

I love that every storage vendor I talk to now is either heavily promoting their analytics capability or gushing about its prominence on their product roadmap. While each vendor does things slightly differently, I think it’s great that there’s been some great progress in the marketplace to extend the conversation beyond speeds and feeds into a more mature conversation around understanding how applications are behaving and what can be done to improve performance to enable improved business operations. QoS doesn’t have to be a super onerous endeavour either. Kaminario have certainly taken an interesting approach to this, and I look forward to seeing how HealthShield develops over the next little while.