Increasingly, on the client side, there are such requirements: “The traditional architectural components of the constructor“ assemble yourself ”like a blade server + SAN + storage + virtualization + add-ons to the cloud are no longer interesting.Innovative solutions are interesting. ”
HP's response to such requests was the emergence of the HP Converged System 200-HC hyperconvergent platform.
HP converged virtualization systems continue to evolve. You can read about last year’s lines in the previous article habrahabr.ru/company/hp/blog/214597 , and this year the line of convergent systems has seriously changed. ')
Let me remind you that such a convergent system according to HP is the basis of the business IT infrastructure, which allows you not to think about the components, but to focus on the result. This level of abstraction is achieved by a set of tested and tested components with quick order and implementation, combined with common management software for work in the customer's environment, with a single center of proactive support, ready for quick changes on demand of the business.
This year’s HP Converged Systems portfolio looks like this:
What is important: all the systems (servers, storage systems and network) are managed by a single software - HP OneView.
By hyperconvergence we mean an even greater simplification of the component base of the system. Tezisno this concept can be described as:
All IT components (calculator, data storage, management) fit in a single “brick” server
The system is linearly expanded in capacity and performance by adding bricks.
The bricks are combined into a cluster for resilience and disaster recovery, the data of one brick is available to all members of the cluster
Minimum time to deploy and build
Many bricks are controlled from a single interface
Figure 1 HP SL2500 Platform
The HP CS200 hyper-convergent system is based on the HP SL2500 hardware platform with super-dense component placement, supporting 2U up to 4 nodes in a single chassis with Intel Xeon E5-2600 v2 processors and up to 24 small form factor disks (6 disks per node).
The SL2500 chassis contains all the necessary redundant components for operation: duplicated power supplies and fans, each node has its own motherboard.
A decorative grille with a lock is installed on the front panel of the server in order to prevent unauthorized removal of disks from the server.
Figure 2 HP Converged System 200 Front and Rear Panel
The software part of the convergent platform includes:
HP OneView InstantON is a deployment and extension manager, a special agent detects when nodes are available for deployment after loading, collects system health statistics, and automates the deployment of vSphere and storage software. This component allows you to deploy a virtual environment in just 15 minutes.
VMware vSphere 5.5, VMware vCenter (optional licenses, you can use your existing ELA licenses)
HP StoreVirtual VSA storage system software - uses disks installed in the chassis of the complex to provide disk resources to virtual machines.
HP OneView for VMware vCenter (pre-installed management system).
Systems with two hardware options are available for order: CS 240 only with SAS disks, and CS 242 with SAS + SSD disks. Each chassis node in these models has the following configuration:
Figure 3 Types of HP Converged System 200 nodes and their specifications
The complex can be ordered as part of a license with VMware, and use the previously purchased license level Ent +. This reduces the cost of the complex.
There are quite a few similar platforms now. What makes the CS200 different from other offers? Mainly - the use of software storage StoreVirtual VSA.
HP StoreVirtual VSA, along with some other embedded tools, provides the system with a level of reliability and performance to the highest corporate standards. All of the following features are included in the basic delivery of the system and do not require additional licensing:
Protection against data failure at the hardware level in the node itself: at each node there is a hardware RAID controller providing protection in case of failure of 1 or more of the 6 disks assigned to the node.
Protection against site failure in real time (Network RAID) protects data when physical components fail, a node fails, the entire rack fails, or power is lost. Converged System 200 nodes can be merged into a network RAID 10 and get fault tolerance at two sites; after restoring a remote site, the virtual machine volumes are re-synchronized automatically. Not only CS200 systems can be added to a network cluster, but also any other servers with HP StoreVirtual VSA software installed.
Data migration between sites with data availability up to 99.999% .
Automatic data tiring at the data block level - often requested data is moved between SSD and SAS levels (for CS 242 model)
Thin provisioning (thin provisioning), integration with VMware and Microsoft file systems is supported, i.e. the system understands when deleted data can be returned as blocks to a pool of free resources (space reclamation technology).
Snapshots are also backed up, and blocks are returned to the pool of free resources after the system has removed the snapshots.
Direct access to nodes that contain target data (Multipathing Extension Module) to reduce the number of hops between nodes when a host accesses writing or reading data and reducing delays.
Access to HP VSA is possible as an object data storage using the REST API protocol.
With the new update StoreVirtual accelerated node load time and increased data replication speed.
The advantages of such a system is that the performance of the complex grows linearly as new nodes are added to the complex, since we add not only capacity, but also computational resources.
The flexibility of this approach is that data can be replicated to any server where a virtual machine with a StoreVirtual VSA license is deployed - including to servers from other manufacturers. Since October 2013, free VSA licenses for 1TB of usable capacity are delivered with any HP server , this allows you to quickly deploy a backup site on any server.
In traditional systems, we will certainly at some stage focus on the performance of storage controllers.
Fig.5. With the addition of HP CS200 nodes to the cluster, linear performance growth is observed compared to dual-controller storage.
Up to 32 nodes (2 clusters of 16 nodes each) of the HP CS200 can be combined under a single control. To merge nodes into a cluster, a SAN network is not needed; a pair of switches with 10GbE support is enough.
Until April 30, there is a special discount on trial sets of CS200 (no more than one set per customer) - ask HP authorized partners. Record of the webinar about the system in Russian: youtu.be/qOzgjdMNwIw What awaits the system in the near future? • Upgrade platform on HP Apollo 2000 system (this spring) - using new processors and memory • Expected HP CS200 certification for MS Lync and Citrix VDI solutions.