Silicon Investor (SI) -- The First Internet Community

STOCKTALK

We've detected that you're using an ad content blocking browser plug-in or feature. Ads provide a critical source of revenue to the continued operation of Silicon Investor. We ask that you disable ad blocking while on Silicon Investor in the best interests of our community. If you are not using an ad blocker but are still receiving this message, make sure your browser's tracking protection is set to the 'standard' level.

Strategies & Market Trends : Gorilla and King Portfolio Candidates -- Ignore unavailable to you. Want to Upgrade?

To: Knight who wrote (37785)	1/11/2001 1:23:23 PM
From: DownSouth	Respond to of 54805

NTAP could go into an existing shop that's running a Fibre Channel SAN, install the NTAP filers and have them run over the existing Fibre Channel "plumbing." No, afraid not. The FC SAN's are laid out in a much more complex manner than what NTAP supports now. Perhaps in the future the existing FC plumbing can be used, but I am not banking on it. Also, the data stored on those SANs are stored in the UNIX and Windows file system format. WAFL has a very different organizational structure, so the data must be "copied" (literally) from its existing media to media formatted and controlled by WAFL in WAFLs data format. EMC FC SAN's require more complexity in their topology because the Symmetrix is attached to the app hosts via channel connect and paths from the app hosts through the Symmetrix to specific data volumes are required. NTAP eliminates the need for physical volumes and paths from app hosts. At the current time, NTAP uses dual channel FC disk drives and controllers and FC-AL so that two filers can have an I/O path to each other's drives in case of a failure of one filer, so the other filer can take over the failed filers I/O duties. This architecture is supplemented by a NUMA connection between the two filer's CPUs so that the non-volatile RAM (cache) devices of the two filers have copies of one another's unwritten records in cache. In the future, with DAFS/VI, FC will be used to build multi-filer clusters sharing disk arrays through FC switches. Also, their (NTAP) filers, I believe, are the only NAS devices supported by all the major RDBMS vendors for "non-direct-attach" access. (Is this still the case, DownSouth?) Yep. NTAP is the only device of any kind supported by any of the big 3 RDBMS vendors for NFS mounts of data. 2. Remote NFS mounts were also undesirable for mission-critical applications because TCP/IP I/O was much too slow and required much more server overhead than direct SCSI I/O. WAFL addresses the performance issue by offloading much of the filesystem processing from the host. WAFL addresses the performance issue in a few ways: a) It is just plain faster on writes than traditional file systems. b) It is a helluva lot faster on writes than traditional file systems with RAID 5. c) It is much more reliable. d) It is easy to attach the filer to the application host via a FDDI ring, dedicated xxBaseT (ethernet) segment, ATM, or some other network architecture, eliminating the competition between "normal" network traffic and file server traffic.(Additional network interfaces can be in the filer so that there are several paths to the filer.) e) The sys admin benefits of using filers for storage are very significant. any idea how fibre-channel-attached NTAP filers' I/O performance/overhead compares to I/O performance of "direct-attach" SCSI-over-fibre-channel storage via traditional filesystems? If you're up for it, take a look at the SPEC SFS benchmarks of SUNW versus those of NTAP. SUNW is running SCSI attached drives, non RAID protected, and the price performance difference is tremendous. There may be some SCSI over FC configs in there also.