Sunday, 11 September 2011
NetApp FAS 2020
Two main reasons for the choice NetAppovih products (with the technical side) are WAFL and replication. WAFL (Write Anywhere File Layout) file system is not in itself, but rather has a role metadatotečnog system over which it may be necessary to make a "normal" system. In this role, we could say remotely compare with VMware VMFS-infection with which is certainly met a number of readers: it is not particularly important that the file system itself, but the important advanced technology that they can implement - and there are WAFL a lot.
In essence, WAFL is a log-based file system which means you can write to it not proceeding according to the logical layout of data on disk (for example, files are always in the same position on the disc), but all logging takes place sequentially (as log) which is constantly growing. In a practical example, two changes in two different files will result in writing to two different places on the disk depending on exactly where the files were first created, but together, one after the other at whatever position on the disc is "the end of the logo," and then only modify the internal file pointers in the position to show the correct location. The consequences of this are numerous: the first obvious thing is to write the new data is done sequentially on the disk, which is significantly faster than random search for the positions (but on the other hand reading is slower for the same reason), and another that is extremely efficient (fast and cheap) place simultaneously copies of all historical data, and snapshot disk contents. To maintain good performance in reading, a special process to periodically monitor usage statistics of data and trying to move or defragment the data so that data which together approach is along the disk (which is significantly faster).
Regarding the log-based mode is an efficient option replication facilities. How WAFL itself must generate information about how the new data replace old parts (or create new data on file), that information is simply sent to other systems, and there applied to existing data.
Together, these features make the basis for the immense popularity NetAppovih devices in a very broad set of customers from Oracle, SAP and NASA to Lidl and Konzum. Even viewed themselves, the possibility is extremely simple and quick and easy snapshot and replication efficiency are top.
NetApp Filer using the same software regardless of the vertical level and complexity of devices: Data ONTAP, which is usually based on FreeBSD, but with major modifications in all layers of the RAID drivers and file systems. The benefit of this strategy over the fragmentation of the products for specific segments of the market (consumer, pro, etc.) is in the simplicity of maintenance and upgrades in case of necessity is trivial to add more equipment or increase capacity.
FAS 2020 is a device to test came configured with 12 SAS disks of 300 GB and 15,000 rpm, with two controllers can be configured in standby and active-active mode. These controllers are almost independent, and its configuration is stored on a basic disk to be used in the system and therefore the two drives (for redundancy) in the basic configuration assigned to each controller for this root volume. The greatest good side of this scheme is the possibility of a robust and easily switch the device into the disc player with automatic detection and reconfiguration. The downside is the loss of four discs, but it can offset the additional purchase of "tray" disc for installation in cabinets with equipment.
Unlike virtually all other manufacturers of similar solutions, NetApp Filer effectively supports only two types of RAID: single and dual parity parity. Type with single parity RAID-4 (which is equivalent to RAID 5, but the information on parity are not broken down by disc), as double is something like RAID 6 but again without the distribution of parity disks, and using a simpler (and faster) algorithm for calculating the parity (and it is called NetApp RAID-DP). For current conditions and capacities of the disks is a dual parity very good idea because it reduces the likelihood of far more data loss than just a single. As for the concrete implementation is concerned, comparing RAID-4 and RAID-5 or RAID-DP and RAID-6, are theoretically "normal variant" (5:06) something more efficient and faster, but NetApp claims that intelligent techniques cacheiranja data in cooperation with WAFL-om resolve potential constraints. What is certain is that this selection of RAID and greatly assists the implementation of the rapid expansion of space to store the new disks.
NetApp rightly has a good reputation among large business customers. Robust replication which works in remote locations (SnapMirror) and the possibility of expanding the space needed to store proved to be highly desirable in today's era of growing and geographically dispersed companies. The question is only - if these are good for smaller companies with smaller requirements and shallow pockets. The answer is related to the assessment of how all the features described actually required of companies and certainly there is a point of profitability, but no matter what we say it is rare that such a solution rich in possibilities.
Posted by mjunta at 23:47