StarWind Enterprise

StarWind Enterprise 5.7

The StarWind Software company which is releasing the best product for creation of iSCSI-storages for your VMware vSphere and Mirosoft Hyper-V servers announced in July, 2011 release of the solution StarWind Enterprise 5.7 which gives even more opportunities for creation of failsafe storages.

This release of StarWind Enterprise 5.7 represents the first step on transfer of the mechanism of fault tolerance of storages to new architecture.

• Improvements of the mechanism of synchronization. Now sending of data between nodes of a failover cluster happens in an asynchronous mode. It means that StarWind Enterprise HA works much quicker now. You can read details of the new mechanism of work of HA in the blog at Konstantin, and we will give the main essence.

Earlier everything worked as follows:

• • When the iSCSI-packet comes to the main node, he sends it to reserve. After the reserve node receives a packet, it sends an iSCSI-command of assurance of receipt of a data unit (ACK), only after receiving which the main node transfers ACK to the initiator and writes a packet on a disk. When iSCSI-commands became much, they were put in queue and there were some delays as constantly it was necessary to make these iterations.

Now everything works so (a certain hybrid of synchronous and asynchronous processes):

• • When the iSCSI-packet comes to the main node, he sends it to reserve. And, without waiting for ACK from the partner, the second packet is sent at once. When queue ends, and the second node will write data on a disk, it sends ACK to the main node which it already transfers to the initiator. So everything works much quicker. Well and protection against data loss at synchronization channel failure is too.

• High availability improvements. There was a management of circuit bandwidth of synchronization (QoS) - now it is possible to regulate prioritizing of traffic at synchronization of HA nodes that allows not to fill all available canal and not to slow down access to the general storage. For this purpose it is necessary to select the item "Sync channel priorities" from a context menu of the HA device. By default for the greatest security the priority costs at clock channel. It is necessary to expose QoS only on the main node, on reserve it will be exposed automatically.
• Optimization of the channel. Now for optimization of the channel several parallel iSCSI-sessions are used. In the next version of StarWind Enterprise 5.8 it is quantity it will be possible to regulate and also setup of number of channels for Heartbeat will be possible.
• Performance Monitoring is a possibility of tracking of performance of a disk subsystem from Management Console (disk transfer rate, average number of I/O operations on targets, CPU and memory load on StarWind server). It looks so:
• The new useful equipment - Snapshot Manager. Now it is possible to manage snapshots through GUI. For this purpose in a context menu for the device it is necessary to select the item "Snapshot Manager".
• The improved Event log. Now at approach of an event there is a notification of the administrator through an icon in System Tray.
• Improvements of GUI. Now Targets and servers can integrate in groups for convenience of management.
• Experimental "deduplication plugin". It allows to set deduplication with sizes of block from 512B to 256KB at choice (instead of 512B now) that allows to increase performance of process of deduplication to ten times, at the same time economy of space on deduplication decreases by only 10-15% (when comparing 512 byte from 4 kb blocks). Besides, load of the CCP considerably will go down and 90% umenshitsya requirements to the amount of OZU.
• ImageFile, DiskBridge. Support of the Mode Sense page 0x3 mode for compatibility with the iSCSI-initiator of Solaris.

StarWind 5.8

• Deduplication - a plug-in for deduplication of storages was processed and now supports journalizing of data. Support of snapshots for the deduplication mechanism is also added.
• The new version of the plug-in providing high availability of storages.
• As failsafe storages raw image files can be used.
• the iSCSI-traffic going on clock channel between the StarWind HA nodes now is transferred more quickly and reliably due to replacement of MS iSCSI transport to own from StarWind (developers claim that in MS iSCSI there are a lot of bugs which remained since 2006 that affects reliability of the solution).
• The channel of signals of availability (Heartbeat channel) is an obligatory configuration for the solution now (in avoidance of inconsistency of data). It is possible to use for it public network - on it there does not pass a lot of traffic.
• The possibility of "Switch partner" - for the existing StarWind HA node can be reassigned a new node (you remember that it is required to update access paths to storages in ESX and Hyper-V).
• Automatic synchronization of nodes after simultaneous switching off of two nodes (for example, power outage). If one of disk devices has more up-to-date data, then they will be automatically synchronized on other node. If you use a technique of caching of write-back and at least one of nodes was switched off incorrectly (hard power off, for example) - that automatic synchronization will not happen, and actions of the user will be expected.
• RAM disk - an algorithm of generation of serial number was changed for respect for its uniqueness within network.
• SPTI - the parameter of alignment of the buffer is used when checking properties of the device during initialization of service.

• The release of StarWind Enterprise 5.7 - is even more opportunities for VMware vSphere and Microsoft Hyper-V storages