Red Hat targets Gluster at EMC / VMware

Red Hat has announced the purchase of Gluster, an open source cloud storage software developer. Gluster software will complement the stack of Red Hat cloud products such as CloudForms and OpenShift, ensuring their flexibility and scalability.

Red Hat buys cloud storage Gluster. What for?

Not so long ago, I wrote about the fact that Red Hat declares war on VMware on the "cloudy" front and now, it seems, its first manifestations.



Among such news about the purchase of Gluster, there are a lot of common words and very little analysis of why Red Hat made this purchase and how it can affect the cloud computing market. I had to think for myself. :-) I collected information from various sources and I hope that I can clarify the situation a little.



How will this purchase affect the cloud market?

To understand why Red Hat bought Gluster , the manufacturer of “cloud storage”, you need to understand why cloud is generally needed. The main reason for the introduction of cloud computing is the desire to reduce IT costs. More precisely, it is effective to use them, i.e. get the overall low cost of computing resources. But in order to achieve low cost of cloud computing, it is necessary that all the components of which the "cloud" is made available at a low price.

')

And today, almost everything to create a “cloud” can be bought relatively cheaply - after all, they are collected from standard x86 servers, and often use Open Source as software. Wait, did I say “almost everything”? When you build your own "cloud", you take x86 servers, network equipment - it is quite cheap. And then? And then you understand that it is necessary to store the processed data somewhere else. Unfortunately, storage systems (hereinafter referred to as “DSS”) today cost tangible money and are a price “bottleneck” that does not allow to fully benefit from cloud computing.



Why is this happening? Because iron producers also want to earn super- profits. And one of the largest players in the storage market - EMC is, quite by chance, the owner of VMware. That same company that dominates the virtualization market. However, IBM, HP and Sun / Oracle also have their own storage systems that need to be sold. So that you can estimate the amounts in question - the cumulative share of EMC, HP and IBM deliveries to the Russian storage market is about 80% ( source ). It turns out that large-scale suppliers of iron cloud computing is beneficial, due to the fact that only they have specialized equipment for large storage systems. Figures from the IDC report :

At the same time, the CAGR indicator of enterprise storage costs in private “clouds” will be 28.9%. By 2015, global spending on storage systems in public and private cloud systems will reach $ 22.6 billion.



According to IDC forecasts, the Russian storage market will continue to grow at a fairly rapid pace, with the external storage segment growing in monetary terms in five years by almost 75%.

For reference, the share of EMC in the Russian storage market of Russia is slightly less than 40%, and VMware, the EMC subsidiary, occupies about 80% of the virtualization market (this is “in the world”, I did not find figures for Russia separately). It's time to recall how in one of my previous posts (dedicated to just the pricing issue), EMC representatives urged to count not separately the cost of services, software and hardware, but only together:

Considering the price of a virtualization platform only apart from the cost of hardware, the entire binding (backup, security, network infrastructure, etc.) and the cost of licenses for application software are devoid of any practical sense.

Anyone who wants to build their own cloud starts by calculating how much it will cost in the end. And if with software, servers and network equipment everything is more or less clear, then with DSS quite the opposite. For example, the price may change significantly if you choose the “wrong” software platform. ;-) And it is precisely the storage system that is today the obstacle that hinders the massive use of cloud computing.



Red Hat representatives have repeatedly stated that virtualization should not be expensive. Therefore, in my opinion, the purchase of Gluster is an attempt to reduce the cost of storing data - the most expensive part of cloud computing. And for sure, EMC / VMware will not like it. ;-)



And what is Gluster?

The company Gluster has become known for its file system - GlusterFS , intended for NAS-storage and distributed as Open Source (GPLv3). GlusterFS is capable of processing arrays with a capacity of up to several petabytes, while the components of the array can be at a considerable distance from each other - they are combined with each other using the Infiniband RDMA or TCP / IP protocols. Gluster is called "the leading provider of horizontally scalable open source storage solutions." One of the staff members said beautifully:

We set a goal to be like Red Hat in the field of storage, and steel storage for Red Hat.

Gluster products fit perfectly into Red Hat's “cloud strategy” and serve as an important element for CloudForms and OpenShift, but beyond that, they open up opportunities for Red Hat to enter the unstructured data management market. It should be noted that although Red Hat does not participate in the OpenStack project, Gluster support for this cloud platform will continue, and the source code of the products will remain open. (However, the Red Hat product code has always been and remains open.)



What about Global File System?

Red Hat already has one development in the field of file systems and storage systems - the Global File System . GFS most often works as part of Red Hat Cluster Suite and allows multiple servers to simultaneously read and write to a single, shared file system on a SAN . Despite the word “global” in the name, this file system is designed primarily for fault-tolerant clusters (HA, High Availability) and, strictly speaking, was developed at a time when no one had ever thought about cloud computing and such large-scale deployments. .



By the way, GFS has an interesting story. The development of this FS began in 1995 at the University of Minnesota. It was written for IRIX by SGI, and in 1998 ported to Linux. At about the same time, the source code of the file system was open. But after the transition (~ 2000) to the development of the company Sistina Software, GFS began to develop as a proprietary product. Independent developers forked the FS, calling the new project OpenGFS, but it did not last long - Red Hat bought Sistina in 2003 and again opened all the code under the GPL license.



What does this purchase mean for Red Hat itself?

Over the past few years, Red Hat’s business focus has changed - if previously open solutions were used on individual business-critical servers, now they are entire data centers. This is especially noticeable on the example of OVA , where more and more service providers and companies directly related to cloud computing are coming.



By the way, the fact that Red Hat recently announced RHEL certification for the Open Compute Project speaks in favor of the “cost reduction for the data center” vector. Let me remind you that the Open Compute Project was launched by Facebook to develop algorithms for efficient and economical scaling of large data centers.



Red Hat has its own virtualization (KVM), IaaS (CloudForms) and PaaS (OpenShift), with the purchase of Gluster will be closed "white spot" in the field of storage. Open source software today is becoming the de facto standard for building cloud computing. All of the largest cloud-hosting work on exactly OpenSource-products (which reduces costs by orders of magnitude), so it is not surprising that Red Hat sees great promise for itself in this market.



I want to believe that just as the emergence of GFS has become an important step for using Linux on critical systems, GlusterFS will also open a new page in the use of Linux in data center and cloud computing. It not only does not have decent “industrial” alternatives (like GFS in its time), but also allows it to drastically reduce the cost of “cloud” data storage.