IP-Tech migrates to Red Hat and realizes enhanced flexibility and scalability

Industry: ICT service provider
Geography: Switzerland
Opportunity: Highly scalable, flexible solution for storage and retrieval of large quantities of data
Solution: Diskless Shared Root Cluster
Hardware: HP ProLiant BL20p Blades, IBM BladeCenter, HP EVA 3000, aiming for full consolidation on Intel chips
Benefits: Open source storage cluster that allows IP-Tech to respond quickly and economically to any demand from its customers; no competition from proprietary vendors.

Thanks to Red Hat Global File System and Diskless Shared Root Cluster by ATIX, we were able to achieve savings of 50% in administrative costs alone. In addition, the new system has a markedly enhanced performance and scalability.

- Donat Grimm, General Manager

Background

IP-Tech AG is one of the largest full service providers in Switzerland. The company offers its customers hosting, voice-over IP, and application services. It was founded in 2000, grew rapidly, and now manages more than 25,000 domains for web and mail services. The company’s IT infrastructure had expanded over the years to more than 30 servers, each with its own operating system and storage device. In view of this heterogenous environment, the indications at IP-Tech were for standardization and consolidations.

Opportunity

The decisive factor for a comprehensive IT restructuring by the full service provider was the project for a recently acquired customer. The standards of our customers are our standards, explains Donat Grimm, General Manager of IP-Tech. We were to operate a media database for a service provider of the digital pre-print stage, which it relied on for producing glossy brochures. This requires not only storing large quantities of data in the terabyte range, but also maintaining constant access to it. We were making no progress with our existing approach, and so we looked for an alternative which would be markedly more efficient, more flexible, and more scalable. We also wanted to reduce costs. It was clear to us that we needed a high-performance cluster solution. In conjunction with HP, Red Hat, and Red Hat Advanced Partner ATIX, we then found the ideal solution for us.

Solution

Starting with the hardware, an entirely new platform was created at IP-Tech based on Red Hat Enterprise Linux. Red Hat Global File System (GFS) forms the core of the new cluster solution for which IP-Tech finally opted. Red Hat GFS is an open source file system which allows several computers in a cluster simultaneous access to the same physical data blocks. ATIX, a Red Hat Advanced Partner for clustering and storage, developed a Diskless Shared Root Cluster on this basis that takes the familiar cluster approach to its logical conclusion and offers IP-Tech a highly flexible, easily scalable system.

When using a rudimentary cluster method, IT services run simultaneously on several servers, and user requests are allocated by a load-distributor. For dynamic services, it is necessary to disconnect the application servers from the data storage and allow them joint access to a storage system. The conventional approach for joint data memory is a file server. This often causes a bottleneck on this server and is a single point of failure of the overall system.

In order to resolve these difficulties with conventional approaches, each server in the cluster should have direct access to the storage devices and be able to read and write to them equally and concurrently. A cluster file system such as Red Hat GFS links in the application servers and a connected storage area network (SAN) and allows parallel file system access by all cluster nodes to a central storage system. The Diskless Shared Root Cluster developed by ATIX takes this approach to the logical extreme and manages entirely without hard disks in the cluster nodes.

In the initial configuration, IP-Tech uses eight HP ProLiant BL20p blades with two Intel processors in the root cluster. We are currently also using other chip architectures, but in the medium term, we want to consolidate on Intel processors. As an industry standard, they definitely offer the best price/performance ratio and maximum freedom in the choice of operating system and applications, explains Grimm. Red Hat Cluster Suite ensures first that the services from a crashed server are transferred to another and second that the load is distributed between cluster nodes.

Red Hat GFS allows parallel access by all cluster nodes to a central storage system constructed from two HP Enterprise Virtual Arrays (EVA) 3000. The GFS Pool Layer virtualizes the storage devices and makes this hardware available independently again. Several devices can be combined in a pool by means of striping or linear connection. Changes in the pool configuration are visible for all cluster servers, and a Volume Manager provides online expandability of the file system. Since many servers in a GFS storage cluster access the same physical data blocks, there is a facility for co-ordinating the distributed access - what is known as lock service. This guarantees data consistency in the file system.

The innovative highlight of the Diskless Shared Root Cluster developed by ATIX is totally abandoning use of hard disks in the HP blades and having these boot directly from the storage system. This configuration is optimally scalable. New resources in the form of new server hardware can simply be added according to the plug & play principle, because the operating system is installed centrally on the storage system. It also enormously simplifies maintenance of the operating system, because in each case, only one instance of the operating system has to be updated.

Benefits

The resulting division of cluster nodes and central disk storage consolidates all information on the structure and content of the cluster in the central storage system. If the server crashes, no information is affected and has to be retrieved. This reduces the retrieval time of cluster nodes to a minimum, as only the server hardware has to simply be replaced to return the system to normal status. This enhances the entire availability of the cluster. The utilities supplied by the EVA storage system are used for back-up. The cluster also allows remote monitoring and management via the ATIX solution, com.oonics GrayHead.

With the Red Hat GFS and the Diskless Shared Root Cluster, we have acquired far more than an optimal system for our new customers, summarizes Donat Grimm. Due to its considerable flexibility and scalability, the system offers the chance to be able to respond quickly and economically to any demand by our customers. There is no competitive system on the proprietary scene. The fact that the solution could only be implemented with Red Hat GFS was the decisive factor, besides the costs, for open source technology. Of course, we are also benefiting from vendor-independence and enhanced performance gained by using Intel platforms.

This entry was posted under Telco, Technology, HP, IBM, EMEA, Global File System. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.