Drive and Rack Scale Storage Architectures

Big data storage problems getting you down? Never fear! James Green put together a video highlighting two companies from last month’s Tech Field Day that are presenting solutions. Igneous offers an array of nanoservers equipped drives, making each network addressable. James also highlights DriveScale’s take on managing big data with their rack adapter to address a pool of JPOD storage. It’s a really great comparison between the two approaches!

Get all the Docker talks from Tech Field Day 12

Docker was one of our presenters for Tech Field Day last month. We were really excited to hear about their latest and greatest. They’ve shared all of the videos from the event on their blog, along with some of the excited reaction. It was great to have Docker at TFD!

Docker is the New Twitter

Rich Stroffolino has an interesting hypothesis here. He sees Docker in a very similar situation to Twitter circa 2011. It’s an extremely popular product, but with an ecosystem of support companies that extend it’s inherent functionality. Twitter decided to subsume more and more of this into what it natively provided, edging out the companies that once complimented it. Will Docker follow the same path? Rich points out why some of their situations are a little different as well.

DriveScale is a new kid on the block with a very seasoned past

DriveScale clearly had the right idea for their Tech Field Day presentation. They led off listing the pedigree of the founding and senior staff, including a deep history with Sun Microsystems, Cisco, and green technology. It certainly made an impact on John White. From there the company laid out their ambitions: give enterprises the configuration flexibility to scale out horizontally in the datacenter. They do this with a 10GbE network adapter to pool a JBOD of storage to traditional pizza box servers with CPU and RAM. This allows storage to be a totally separate concern for scale. John also liked their strategy of initially targeting Hadoop as a primary use case. It’s not a huge market, but definitely one they could become well known within, given the strength of their solution.

Cloud Storage? In my DC? Yes please! Enter Igneous

John White reviews what he saw from Igneous Systems at Tech Field Day last month. Overall, while cloud storage in your datacenter isn’t necessarily groundbreaking, he likes the approach that Igneous is taking. He particularly calls out the RatioPerfect Architecture, which puts compute on each drive. This eliminates a lot of I/O bottlenecks. John found the price competitive, especially compared to AWS. Add in the latency benefits on on-site storage, and Igneous has a compelling solution.

Decouple Disks and Compute with DriveScale

Eric Shanks takes a look at what DriveScale presented at Tech Field Day last month. In their solution, he sees a real value play for Hadoop workloads. Whereas other applications can used virtualized storage arrays, Hadoop benefits from direct drive access to their distributed file system so it can manage storage. DriveScale allows for this with their disaggregated storage solution via their adapter. This allows you to add storage without throwing a whole other pizza box into the rack. It’s a pretty specific use case, but Eric sees it giving a lot of value to the growing base of customers using Hadoop.

Efficient Resource Use Takes Interesting Turn at Scale

Tim Miller has some thoughts about DriveScale, which he saw at Tech Field Day last month. But to fully understand their solution, he delves back into a little bit of IT history. The brief but informative look back shows how IT has moved to the Big Data mindset of building clusters for each application. In the foreseeable future, we know that compute will surpass these Big Data setups, resulting in inefficiency. Tim thinks DriveScales disaggregated storage solution is setup for this reality. On a practical level, he really liked that DriveScale’s solution doesn’t insert themselves into your storage supply chain, relying instead on a simple rack adapter and software. Overall it gives a great perspective on where DriveScale is going in the future.

Transform and Scale Out with Isilon

Rich Stroffolino gives a rundown of the Dell EMC presentation from Tech Field Day last month. They highlighted their latest hard disk offerings for their Isilon platform. This includes a look at the history of the platform, their latest node offerings, and their hybrid flash solution. Sadly, he was not treated to a look at their new all flash Nitro array. Still, Rich offers some interesting insight on how the scale of Dell EMC differed from some of the emerging vendors he also saw at Tech Field Day.

The Atlas File System – The foundation of the Rubrik Platform

Rubrik’s presentation seems to be impressing a lot of people. One of their biggest features is their ability to distribute backup data across various rack mounted “Brik” appliances, with easy scale out provisioning. Mike Preston has a writeup going into some of the secret sauce that let’s Rubrik do this: the Atlas File System. One of the most interesting features is how the system is able to deal with failure. The file system is able to do replication on both a node or drive level. This allows the overall system to have a whole node or up to two drives fail without any danger of data loss. Mike goes into a deep dive on all of the technical details, so make sure to check it out.

Forward Thinking Backups

Rich Stroffolino takes a look at what Rubrik presented at Tech Field Day earlier this month. They take a extremely focused approach to backups. Their solution allows for backups to their various rack mounted “Brik” devices from various sites with simple scale out and management, and even allow for seamless flow over to S3-compliant cloud storage. While not as expansive as other vendors, the focus of Rubrik makes them stand out.

The Igneous Synthesis

Igneous Systems proposes to offer a storage appliance that will allow you to get the benefits of Infrastructure as a Service while keep all your storage local. Rich Stroffolino gave their Tech Field Day presentation was impressed with how the company was able to synthesize the two aspects. As he points out, this often isn’t an easy task, but Igneous gives you robust local protection in their all in one storage device, while giving you cloud-centric fleet management of the entire device network across all customers.

Scale-Out. Distributed. Whatever the Name, it’s the Future of Computing

Alex Galbraith was inspired to write about the wonders of synthesis. The combining of established ideas to create something new drives a lot of innovation, and what Alex saw from Igneous Systems is a prime example. Their prime innovation, taking the abundance of horizontally scaling compute power, and putting a processor on each drive. This effectively reverses the typical storage scenario of have a small number of large fault domains. With Igneous, the fault domain is exactly one drive, making each failure negligible.

Scale-Out Storage Through Disaggregation With DriveScale

Ethan Banks took a look at DriveScale’s disaggregated storage solution at Tech Field Day this month. Their overall strength relies on their flexibility. DriveScale makes it both easy to manage a true scale-out solution, while also providing potential savings down the upgrade path. They do this by separating storage from compute, so while the initial install they envision being cost-neutral, down the upgrade path, you don’t have to pay for storage you already have. Ethan’s heard similar “it pays for itself” pitches before, but seemed to think the DriveScale solution could actually deliver on that promise.

First Look at Cohesity Cloud Edition

Matt Crape follows up on his Cohesity primer now that he attended Tech Field Day. The big thing that he saw at their presentation was their introduction of the Cohesity Cloud Edition, which effectively lets you take their secondary storage solution and spin up an appliance in the cloud. Matt’s only compliant: Cohesity’s solution is so robust and feature complete, it seems like it would be a good way to handle primary storage!

Igneous – On Premises, Cloud Managed, Scale-Out Storage

Ethan Banks gives an overview of what Igneous Systems presented at Tech Field Day this month. It’s an interesting solution, while acknowledging the plethora of open source options for developing a storage array out there, the Igneous team walked the delegates through why they developed their own data path and hardware architecture. Ethan digs into how the company deals with drive failure, their secret he dubs “the wide Igneous stripe” , a 20+8 layout scheme.

On-prem Cloud Storage with Igneous Systems

On-prem cloud storage? Sounds like a contradiction in terms. But much like jumbo shrimp, Igneous Systems makes it work. They presented at this months Tech Field Day, and Matt Crape was intrigued by what he saw. Igneous is offering an end-to-end solution with their storage array. Matt liked how Igneous rethought their solution from ground zero. Effectively, previous storage arrays are dependent on a few SAS cards, which provides a big bottleneck if and when those cards go down. Igneous calls their architecture RatioPerfect, they effectively put an ARM interposer on each drive, taking the fault domain from a slew of drives to just one. Matt still has some questions about practical implementations of their solution, but on an architectural level, it’s clear Igneous did their homework.

Scaling storage with an army of ARM!

Mike Preston really seemed impressed by Igneous’ storage solution presented at Tech Field Day. They offer a really interesting solution to storage. Instead of have multiple drives beholden to a single powerful Xeon CPU, Igneous proposes their RatioPerfect system, one CPU for each drive! They do this with an Ethernet equipped ARM board attached to each drive, what Igneous calls “nanoservers”. These are then put together in a JBOD within a 4U rack. On top of that, Igneous proposes their solution as end-to-end, the customer hooks it up to the network and Igneous takes care of the rest of the management. Give the rest of the piece a look, Matt gets into a lot of detail about their management plane, which he thinks might be the key to future success.

The Container Storage Persistence Challenge

James Green published a great video summarizing the current state of persistent container storage. One of the more interesting offerings was presented at Tech Field Day this month by StorageOS, which runs on the application layer as a container itself. If you need to get up to speed on container storage, James made it easy with this video.

Improving Stateful Container Storage with StorageOS

Ethan Banks lays out some of the problems with containers. While originally envisioned for application development, they’ve quickly worked their way into infrastructure and operations. With their easy fluidity and reduced requirements, it’s easy to see the benefits of this containerized approach. But as their role has expanded, their deficiencies have become more profound. One of the major issues, containers are generally stateless, but as they expand further into other IT sectors, the need to map these to storage volumes become all the more glaring. StorageOS provides a solution to this problem by making it easy to manage the underlying storage of these containers. They do this by running at the application layer as a 40MB container, with tight integration with Docker, Swarm, and Kubernetes. Click through to Ethan’s piece for a complete breakdown of how StorageOS works, and how you can try it out for yourself.

The Silent Threat of Dark Data

How much do you think about secondary storage? Probably not a ton. It’s all the data that’s not mission critical, the reams of backup data, archives, test/dev, and machine generated data that lives in separate silos. Cohesity demoed a holistic solution to deal with this mess, but why do you need it? James Green knows the answer: dark data. All that data that sits in storage, without metadata or context, data you don’t even know is there. Its a problem for any business that keep financial or medical records. All it takes is one malicious attack, and all that data you didn’t even know you had becomes a huge liability. Cohesity’s solution seems like a ray of light into this work. James breaks down how exactly they go about it.