The Silicon Corner

This is only my second post with The Silicon Corner, but I think it’s an important one. A number of you have had your eyes on this blog since Stephan Michard created it, because it represents a powerful idea. When we unite, we become stronger.

Stephan brought together a group of people from completely different backgrounds with completely different ideas on how a blog should be run. At no point did he tell us what to write, he simply asked us to write. I’ve made several attempts at running a blog in the past, but my first post on The Silicon Corner was more successful than any blog I’ve ever written.

Setting the stage

I’m sitting on my couch at 11:00PM. It’s the night before the start of EMC World — one of the biggest conferences in the Information Technology industry. Over the next week, we will launch a number of new products. Each with its own unique strengths. Each another foot forward in another direction the industry is heading.

We’ve been talking about a new product internally for the past few months that we think you all will be really impressed with. It’s got a slick new HTML5 interface, it comes in all-flash and hybrid flavors, and it’s got some really cool innovation under the hood.

I’m trying to decide how to communicate just how simple, affordable, and flexible this new product is. At the same time, I want to recapture the success of my last blog by replicating the same formula. This involves scanning through slide decks, noting down bullet points, highlighting differentiators, and screen-shotting diagrams. But I’m beginning to realize that this formula won’t do this product justice.

This product truly is everything I’ve stated and more — I’m not retracting on that. But when EMC marketing named this product “Unity” they tapped into a concept much more powerful than any of the messaging I’ve seen around it.

The official theme of EMC World 2016 is “Modernize”. The idea is that the modern data center rests on 5 pillars: 

flash for primary storage

cloud-enablement

the ability to scale-out

intelligence held by software, not hardware

having an infrastructure you can trust

In order for you to innovate, you have to optimize. In order for you to optimize, you have to modernize your IT infrastructure, automate your service delivery, and transform your people and process. I stole this message nearly word-for-word from here.

It’s a great message and I believe it whole-heartedly. But due to a number of external factors, an undercurrent is beginning to surface. It’s a message that’s not about our infrastructure, but about us. Unity (the product) aligns to this message from a few perspectives, but as we zoom out, you’ll see this message is resonating through EMC and the industry as a whole.

When we unite, we become stronger.

I started with EMC one year ago (to the week). The first challenge I encountered was describing to my friends and family what exactly my new company does. A lot of people have tried to tell me that EMC is a storage company — and that’s partially true. But if you check our corporate profile, you’ll see that “cloud computing” pops up in the 2nd sentence, and “storage” only appears in the footer.

In my year at EMC I’ve learned more about cloud than about storage. There’s public, private, IaaS, PaaS, SaaS, automation engines, schedulers, service catalogs, and more. And looking at cloud from a technical perspective you can see an over-arching trend. Customers want public AND private. Customers want VMs AND containers. Nothing is an either/or question any more. There is certainly an aspect of unity to this.

But when I talk to customers, they aren’t as concerned about the technical implications of cloud, they’re concerned about this concept of a “cloud team”.

“What if a VM gets assigned to the wrong VLAN?”

“What if an app owner requests more capacity than they need?”

“What if someone pushes an `rm -rf`to all of our Linux systems?”

These aren’t questions that are asked by a team that is united. In order to successfully run a cloud infrastructure, all infrastructure roles have to come together, acknowledge each other’s strengths and weaknesses, and bring one another up to speed. Because when we unite, we become stronger.

But it’s not just separate infrastructure teams that are merging together. Developers also utilize cloud infrastructure to create cloud-native applications with devops toolchains. In the same way that network teams have to open up communication with virtualization teams, development teams now have to open up communications with infrastructure teams.

The devops community has really nailed this concept by framing the discussion around empathy. Gone are the days where you can push code to production at 4:45 PM on Friday, pack up, and go home. Gone are the days when sysadmins can say “the lights are green, it must be an application issue”. Throwing problems over the wall doesn’t benefit anyone, because when we unite we become stronger.

EMC and Unity

While these trends are slowly taking hold at a macro level, I see them accelerating rapidly within EMC. What was once a storage company is now a united federation of companies with solutions at every level of the stack. Our motto used to be “Where information lives”. I believe that phrase applies even better today because information lives everywhere, be it sitting on a hard drive, distributed across a computing cluster, embedded within a container, or getting filtered by a firewall.

When EMC united with VMware, VCE, Pivotal, Virtustream and RSA, it was with the understanding that each company had its own area of specialty and would be allowed to operate freely within that area of specialty. But each of these companies are still united behind a common strategy because when we unite, we become stronger.

Another great example is VCE, our converged systems division. VCE created the converged infrastructure industry and is the dominant leader in that category. At the core of VCE’s success is a longstanding partnership between EMC, Cisco, and VMware. When three companies who are the best at what they do unite to sell one product, everybody wins. Because… well you guessed it… unity.

Disclaimer: VCE now sells several products, and they are all equally awesome. Unity (the product) will be available inside a VCE Vblock from day one.

With all that said, there’s one big question that is on everyone’s mind at EMC World 2016. What will happen when Dell and EMC merge to become one new company? Our competitors have chosen different strategies. Either to divide and conquer, or to keep another company at an arms length through OEM relationships. No one can say what the right strategy is, but we are choosing to stick to our guns and strengthen ourselves by uniting with Dell. Because when we unite, we become stronger.

Bringing it back to the product launch.

It’s not normal for me to write so much about feelings and concepts. But before I launch into speeds and feeds, I want to call out one more example. The VNX story is a story of unity. 

EMC acquired Data General in mid 1999, and two companies became one. At first, we struggled to successfully position two previously competing products — Symmetrix and CLARiiON. Eventually, EMC emerged on the other side of the dotcom bubble with the leading products in mid-range and high-end storage.

As the NAS revolution took hold, we took CLARiiON one step further by uniting it with Celerra. And thus the VNX was born. By putting file-serving data movers in front of our block system, we created a stronger solution that empowered customers to deploy LUNs and filesystems in the same cabinet.

A lot of time has passed since then, and the VNX has aged like a fine wine. However, the time has come once again to introduce unity into our mid-range system. This time, in the form of an architecture rewrite, and a new family of systems.

Unity is a brand new mid-range unified storage product. It’s built on Linux, not Windows, so it’s completely fresh code. But it retains a lot of the VNX functionality you’ve come to know and love. 

We’ve committed to all-flash for primary storage, but the option for spinning disks is still there. FAST VP and FAST Cache allow different drive types to operate together in a sort of united harmony. I believe the price-point for all-flash will be so compelling that few will go the hybrid route. But even in all flash configurations, FAST is still working behind the scenes to coordinate data placement on TLC, MLC, and SLC flash drives.

It’s not two systems in one cabinet — there’s no separate file hardware. In fact, unity (the concept) is so prevalent in Unity (the product) that LUNs, filesystems, and VVOLs can all be deployed directly from the same pool. There’s one replication method and one snapshot mechanism for all data types. I’ve been known to confuse SnapView and VNX Snapshots as well as MirrorView and Replicator. You don’t have to worry about that any more, it’s Unity.

The new Unisphere interface is smooth and well organized. But the best part is that it just works in the browser. No separate Java client. Filesystems are now based on UFS64 and can scale much larger. The system vaults to M.2 flash drives on the storage processor, which opens up a lot of space on the system drives. Because EMC remains united with our federation partners, Unity can be consumed as a purpose-built appliance, a virtual machine, or a fully converged system.

At launch, there will be four different models that fit in where the VNX 5200–5800 used to. These stats are temporary, as the disks will be getting a lot bigger and more dense the near future.

300(F)

480 TB raw capacity

2 x 6 core, 1.6 GHz CPU

48 GB RAM

400(F)

800 TB raw capacity

2 x 8 core, 2.4 GHz CPU

96 GB RAM

500(F)

1120 TB raw capacity

2 x 10 core, 2.6 GHz

128 GB RAM

600(F)

1600 TB raw capacity

2 x 12 core, 2.5 GHz

256 GB RAM

And that’s the full spiel. Unity is more than just a new mid-range storage system. It’s a concept that’s prevalent to us as engineers, as a company, and as an industry. We’re thinking about unity when we talk about federation partners and converged systems. Unity is on our mind when we discuss industry trends. And unity will be the primary factor when we start our new chapter with Dell in a few short months.

I am happy to annouce my friend and colleague Steven Mackay as a new member of the crew. In Steven’s first blogpost “Somewhere between clouds …” he nicely summarized the different cloud solutions from EMC. Steven, Richard and I met during our on boarding process at EMC. During these two months we worked closely together for our assignments and teamed up for our final exam. I get to know Steven as a real team player, being focused on the team’s success by finding the best possible solution for the customer scenario. I admire his positive attitude and his sense for elegant professional outfits ;-)

As a proud Scotsman, Steve resides in a small town south of Edinburgh. He attended Heriot Watt University to attain a BSc in Computing for Industry and joined EMC in 2015 as a Presales Systems Engineer. Steve has a background in infrastructure and operations, with experience in VMware, Cisco and Microsoft to name a few. 

A self-proclaimed podcast junkie, Steve keeps up-to-date by listening to podcasts such as The Hot Aisle, The Cloudcast (.net), In Tech We Trust and GreyBeards on Storage.

Fascinated as to how modern applications are being deployed, Steve is keen to absorb as much information as possible on Cloud Native Apps and the platforms they will run on.

Likewise, how the application of both Flash and HCI in today’s Datacentre can have substantial impact on IT operations and business outcomes.

Steve is a member of EMC Elect 2016.

by Steve Mackay

Is there a more disruptive technology concept at present, than cloud?  Personally, I don’t think so. There is nothing new about it but it has had, is having and will continue to have sizable influence on the technology sector, business outcomes and even our culture.  10 years ago, had someone remarked they were saving photos to ‘the cloud’, I’d have looked at them like they had two heads.

From a business perspective, identifying an optimal technology operational model, invariably involves some form of cloud computing. Be that operating a traditional datacentre with minimal cloud services; to an organisation with an entirely private cloud-like infrastructure, complete with key cloud principles such as self-service, automation, orchestration and broad network access; or the other end of the spectrum - running everything in a public cloud.

Advocates of public cloud will argue that large scale cloud providers offer both economic and flexibility advantages due to massive economies of scale and ever expanding service catalogues. Whilst on the other side of that fence, it is argued that public cloud does not offer a better economic model for perpetual, committed, traditional mission critical workloads over the long term. The recent news of Apple’s ‘Project McQueen’ and Uber investing in Datacentres rather than running in the cloud, could offer evidence to this point. Throw in data sovereignty for good measure, combined with the potential for reduced visibility and control, and then it is clear that one size never fits all.

The reality is that the majority of businesses today find a balance by using a mix of traditional, private and public elements. The EMC Federation offers products and services to enable the use of cloud in many shapes and sizes. From a full hybrid cloud platform; to a structured PaaS offering, which can reside on private or public cloud infrastructure; or simply tools which allow data placement or protection across different types of clouds.

The following list is by no means a definitive catalogue of every single EMC Federation product and service pertaining to cloud, but simply highlights some of areas where EMC enables customer to reach that optimal operational state.

CloudArray

It makes sense to start by talking about CloudArray. Essentially a cloud ‘gateway’, this physical or virtual appliance presents a traditional storage target (CIFS/SMB, NFS, iSCSI) to the internal network, whilst behind the curtain is actually persisting data to the cloud. This ability to burst out to cost effective cloud capacity is bolstered by the fact that it can plug into various object storage platforms, be that cloud storage built on EMC ECS or EMC Atmos, or common cloud providers such as VMware vCloud Air, Virtustream, AWS or Rackspace (and many more). To appease security concerns, CloudArray encrypts both at-rest and in-flight. Example use cases could be backup or archive applications pointing at a target presented from CloudArray, or simply just a Windows share presented to the internal network for use as a general storage area for tier 3 data.

 The back of a napkin explanation:

CloudBoost

CloudBoost provides cloud enablement for EMC Data Protection products Networker and Avamar. Offering an alternative to tape, CloudBoost extends Networker’s long term retention capabilities to the cloud. Like CloudArray, connectivity to both private and public is supported; as is at-rest and in-flight security; with the added benefit that source side deduplication is also available, which can reduce the size of the network connection required to get the backup data from A to B.

CloudBoost explained:

CloudPools

Gone are the days where the cost of disk resulted in additional scrutiny on which data to retain. Today, is anything ever deleted? Unstructured data is growing and growing exponentially. EMC Isilon is geared up for the challenge of huge unstructured data growth, due to its node based, scale out architecture. Whilst Isilon can scale to capacities of 50PB, it also has a native and automatic cloud tiering technology called CloudPools. CloudPools expands on the SmartPools functionality, which can tier data across different node types within the cluster based on a wide range of policies. Even though the data resides on the cloud, CloudPools maintains a single namespace across the internal Isilon cluster and the cloud tier, so that for all intents and purposes the end user accesses the data in the exact same way they would if it was stored locally on the Isilon cluster. With private and public cloud support, in addition to encryption and compression options, CloudPools arms customers with seamless and pain-free ability in managing data growth.

Back of napkin version:

Spanning

Spanning is the industry leader in SaaS protection (e.g. SFDC, Office365). ‘Why do I need to protect my SaaS applications? They are in the cloud!’ True, SaaS providers will provide resiliency and redundancy to keep things up and running but that won’t protect from user errors, sync errors and malicious intent. Additionally many compliance standards are requiring that SaaS data be backed up in the same way as on-premises application data, furthering the case for Spanning. With utilities that can integrate into the SaaS interface and mobile companion apps, Spanning is also simple to operate.

Spanning at a high level:

Cloud Native Applications

Taking a slightly different track and looking at it from a the viewpoint of deploying modern web scale applications for web and mobile consumption, Federation companies Pivotal, VMware and EMC all offer technology to empower the developer.

Pivotal Cloud Foundry for example is a structured Platform-as-a-Service solution allowing customers to focus on building applications and not concern themselves with the underlying infrastructure. PCF will happily reside on top of public or private cloud infrastructure and in the case of public cloud, implementing PCF can help avoid the potential lock-in that can sometime be associated with using native PaaS tools. Programming languages and their associated frameworks such as Ruby (on Rails), Java (on Spring) and Node.js (on Express) are all available. These options give developers the flexibility and tools required to be able to build apps, provision them, scale them, and also introduce automation and continuous delivery.

Pivotal Cloud Foundry:

Continuing on the theme of containerised applications, VMware are cognisant that application design is changing. Scalable, autonomous, resilient micro services provide a more suitable platform to host cloud native applications. Realising that whilst ESXi provides an established and feature rich platform for traditional applications, due to scalability limitations it is not optimal (in some cases) for cloud native applications. VMware Integrated Containers and the VMware Photon Platform have been developed with large scale distributed containerised apps in mind.

VMware Integrated Containers is built on the tried and tested ESXi platform but allows the container to live in a lightweight version of a VM. This provides the bridge between traditional apps and cloud native apps, allowing both application models to live on the same stack, with both having access to the same set of features, security etc.

Alternatively a greenfield site with no ESXi estate may be best suited to using the Photon platform. The specifics of Photon are ingredients for another blog post but the key is that the platform provides a prime infrastructure layer for deploying containerised applications in the same vein that ESXi provides an infrastructure layer for deploying traditional applications.

Have a look at Integrated Containers & Photon:

Natively, Docker containers do not have a persistent storage layer to save data. What’s more, not all apps are of an ephemeral nature, some require the ability to harden to storage. Developed by the team at EMC{code}, RexRay serves as a storage orchestration layer for Docker, in order to provide persistency for apps that need it. RexRay is abstracted from the underlying storage platform, essentially acting as a driver between Docker and the storage (similar to Cinder in OpenStack). To muddy the waters on that analogy, RexRay can actually plug in to Cinder (so like 2 layers of abstraction from the storage) and other familiar platforms such as XtremIO, ScaleIO, Isilon and AWS to name a few.  RexRay also has further support for Mesos frameworks and is currently available on GitHub.

RexRay:

Summary

The challenge in business today is finding the harmony between operational flexibility, value for money, control and compliance. Utilising a mix of traditional and cloud is usually the key to finding that perfect blend. This post focused on two things; firstly on some of the EMC products and services that can enable companies to find that balance; and secondly, touched on some of the more progressive architectures and tools that can be utilised to build and deploy cloud native applications.

Other products and services that get a notable mention are Virtustream, Federation Enterprise Hybrid Cloud, multiple products across the VMware portfolio and also, (as reported on The Register and Chad Sakac’s VirtualGeek) a yet-to-be-released product from EMC, which will provide a turnkey OpenStack solution (hardware and software), with the end game of being able to serve as a platform capable of deploying and managing containerised workloads both optimally and at scale.

I am happy to annouce my friend and colleague Richard White as a new author on the siliconcorner. You may have read his first awesome article on EMC’s DSSD announcement last week. Richard and I met during our on boarding process at EMC. After spending two month together at the new hire bootcamp both of us transitioned into our field roles. The two of us try to stay connected through Twitter, Slack or (oldschool) mail as much as it is possible with 5000 miles apart.

Richard writes from his home in Lewisville, TX. He's been interested in Information Technology from a young age, always wanting to follow in his father's footsteps. Between tearing apart old PCs and tinkering with stylesheets in the blogs of yester-year, Richard always found himself in the nuts and bolts discovering how technology works.

Richard studied Information Technology at Tarrant County College, obtaining his Associates Degree along with skills in storage, virtualization, networking, servers, operating systems, security, scripting, and more. Since graduation, Richard has taken a light course load alongside full time work as a customer and a reseller. In 2015, Richard joined EMC as a Systems Engineer, and he hasn't looked back.

You can expect to see Richard write about a wide range of topics, but lately he's focused on cloud technology, devops, and analytics. He sees these fields as new frontiers, yet to be explored by the masses.

Today (02/29), EMC announced general availability of a couple new products during our #QuantumLeap event. This includes VMAX All-Flash and the DSSD D5. Both of these new products have huge implications for new opportunities arising within the modern data center. 

Since I’ve already spotted quite a few write-ups on the VMAX All-Flash, I’d like to use this post to focus on the DSSD D5 - it’s application, advantages, and architecture.

The first thing to understand about DSSD is that it doesn’t fit into any of the traditional swim lanes we’re used to seeing in infrastructure hardware. It’s true that DSSD is an all-flash array. However, it would be a mistake to group DSSD in with XtremIO and its competitors.

The primary use cases for DSSD are:

High performance databases & data warehouses (Oracle, SAP)

High performance apps running on Hadoop (Cloudera)

High performance custom applications (SAS, GPFS)

Notice the pattern? DSSD shares characteristics with transactional block arrays, NAS clusters, and web-scale object systems. But in each case, the differentiator is extremely high performance.

In order to achieve this level of performance, organizations have historically attempted one of two things:

Replace spinning disks with flash. - While this is effective, data is still subject to the latency caused by the network fabric and I/O stacks.

Install flash directly into servers (What I like to call “DAS Flash”). This method also has benefits, but you lose the efficiency that comes with having shared storage (accessibility, centralized management, etc) and expose your organization to the dreaded “silo” effect.

DSSD addresses both of these problems by placing shared flash storage into the same rack as the server, connected to the same PCI bus as the server’s CPU. This is accomplished by using the NVMe protocol to create a PCI mesh fabric that connects DSSD to as many as 48 servers.

With this setup, the flash storage is still centralized into one location, but it now has a direct connection into the CPU running the application. Network latency and IO bottlenecks are no longer a concern, and you retain the benefits of shared storage.

The image above shows a couple of the hardware components. On the left, you see the 5U DSSD D5 (not quite a palindrome) chassis which holds 36 specialized “flash modules”, I/O modules with 12 PCIe hubs each, and redundant swappable control modules.

On the right, you see connectors going into a client card. This card and its respective driver allow DSSD to bypass the entire stack and speak directly to the application.

The flash module:

The I/O module:

The control module:

The client card:

In a traditional architectures, there is a speed degradation of three orders of magnitude between the memory bus and the storage array. DSSD’s ability to bypass the kernel closes the gap between memory and flash, and in some cases surpasses the speed of DRAM.

I’m sure that last sentence got your attention, so let’s talk numbers. Quantitatively, the DSSD is capable of pushing 10 million IOs per second, with sub 100 microsecond (not milli) latency, at 100 gigabytes (not bits) per second on 100 terabytes of flash storage.

The Twitter teasers weren’t lying. We told you it was fast. #FasterThan anything anyone has ever seen.

Moving back into the weeds, let’s talk about software. Most all-flash arrays have block-only access - maybe one or two startups have started experimenting with file. Believe it or not, DSSD is an object system. That’s not to say you can’t store file shares or raw block data - you absolutely can. They just look like objects to the system.

As you can see this is truly a unified system, but again the use case leans heavily to the high performance side. It may store files and objects, but this is not a place to put your cold archive or home directories. The exception is when you integrate DSSD into your data lake strategy where it compliments Isilon and ECS very well. 

In order to access these objects, there are three different options:

Flood Block Driver for high performance databases that crunch raw data.

Libflood Direct API Access for customized apps where developers want to speak to the array directly.

API interface plugins for applications running on HDFS, namely Cloudera. Pivotal and Hortonworks should have plugins shortly.

There’s also some really cool innovation around data protection, and maintaining high performance despite it. Between features like space-time garbage collection, flash physics control, and cubic RAID, DSSD truly has enterprise-level reliability.

DSSD is constantly monitoring the individual flash dies to determine best data placement. By keeping tabs on environmental factors like temperature and by decreasing write amplification, DSSD can make its flash modules last 2x longer than other systems. 120PB can be written to these modules over their lifetime. 

What DSSD doesn’t have is services like deduplication, compression, snapshots and replication. These features add performance overhead, and in situations where these features are required - XtremIO is probably a better fit.

Today VCE, the converged platform division of EMC, announced its new hyper-converged appliance VxRail. EMC’s portfolio of converged infrastructure consists of Blocks, Racks and Appliances. Starting on the one side with VBlocks and VxBlocks, over VxRack, and towards VxRail Appliances. The advantage of such a portfolio is to have the best fit for every workload and to meet differing customer needs.

VBlock systems are an example of traditional converged infrastructure comprised of dedicated elements for compute, storage, and network. On the other end of the spectrum, a hyper-converged infrastructure is a software defined architecture with integrated compute, networking, virtualization, and software-defined storage. The fact of having a fully virtualized environment enables to decouple compute, network and storage function from the underlying hardware, and to run on industry-standard x86 hardware. VxRail is built for branch offices, small to medium enterprise data centers and distributed enterprises. Companies that follow the route of hyper-converged infrastructure can start with a small cluster and scale according to their business needs.

A VxRail appliance consists of 4 nodes. Appliances are available in different flavors with respect to the number of cores, the amount of RAM, size and type of storage medium. A cluster can scale up to 64 nodes in total, which corresponds to 16 appliances. Within a cluster, it is possible to mix different types of appliances.

Being fully virtualized, VxRail has a deep integration with VMware. VSAN 6.1 is integrated at the kernel level, delivering the software defined storage layer. VSAN itself is managed with vSphere 6.0, enabling high performance, automated scaling, and guaranteeing shortest IO paths. VSAN 6.2, as well as the possibility to scale in 1 node increments, will be available in Q2 of 2016.

VxRail is the way to have an IT environment out of the box and up and running within minutes. To summarize: don’t build what you can buy, simplify your life and concentrate on your core business.

Last week on Monday, EMC announced the participants of this year’s EMC Elect program. This is a recognition program for EMC advocates (details can be found here). Somehow, I was selected to be part of the class of 2016.

 I knew that my GSAP fellow, and good friend, Richard White (@RichardWhiteIO) nominated me for the class of 2016, late last year. I didn't really think that I would have a chance to be a part of this elected circle. Sure, I like to tweet and blog about my day to day job as a Systems Engineer at EMC, and about our industry leading products and solutions, as well as some nerdy stuff. However, my online activities were highly unstructured and didn't follow a strict plan or strategy so far. Therefore, I was very surprised and overwhelmed when I got the news.

On the one hand, I am very excited to be a part of the EMC Elect class of 2016. On the other hand, I am extremely humble. Many of fellow EMC Elect colleagues are industry renowned experts, who publish very cool and interesting content on a regular basis. I consider many of them being my idols, in one respect or another.

I don't want to make false promises about blogging frequency or future content of my blog post, but I know that EMC's pipeline of new products is packed with tons of exciting stuff. One very cool benefit of being an EMC Elect member is that you get early and exclusive technology updates about the coming announcements. I want to take the advantage of this benefit and grant the readers of my blog an access to interesting insights of our technology stack.

So, stay tuned for the cool stuff …

This week EMC announced several new features for its Isilon product line. Isilon is EMC’s scale-out NAS solution. Each Isilon cluster scales from 3 to a maximum of 144 nodes. Isilon runs one operating system called OneFS which allows all nodes in one cluster to act as one file system. With intelligent tiering algorithms, workloads and data can be spread across the processing units and drives of several nodes. Hence distributing the load over all nodes in one cluster avoiding any bottlenecks within the cluster.

This week the following features were announced:

OneFS.Next – is the next release of Isilons OneFS operating system. From rolling updates to non-disruptive updates. With the new operating system all future upgrades can be performed without any interruption for the users of the Isilon cluster.

IsilonSD Edge – is a fully software defined storage solution for the Isilon product line. It is designed to be deployed on commodity hardware for a capacity up to 36 TB. It is targeted for the use in remote offices in order to extend the edge of a company’s data lake. As many other products from EMC’s SDS portfolio this software will be available free and frictionless.

Isilon Cloud Pools – is the capability to extend a Isilon cluster seamlessly to public, private, and hybrid clouds. Data which is not accessed frequently can be transferred to the cloud. Hence the Cloud Pools feature enables to use the cloud as an archive tier delivering unlimited capacity and compelling economics.

To summarize, the features presented this week will offer non-disruptive updates and enable to extend the edge of a company’s data lake to remote-locations and to the cloud.

Find out more on the official EMC website.

Nice example how a traditional company re-invents itself by adopting an agile way of working. The agile methods are not generally applied everywhere, but everywhere where it makes sense.

Tolles Ergebnis: