NetApp filer

From Wikipedia, the free encyclopedia

In computer storage, NetApp filer, known also as NetApp Fabric-Attached Storage (FAS), is a NetApp's pioneering network attached storage (NAS) device. Filer is essentially enterprise-class storage appliance – a kind of computer appliance which sole purpose is serving the storage over the network using NFS, CIFS, FTP, iSCSI, Fibre Channel (FC), or similar protocols.^[1] It is equipped with large disk arrays.

Most other large storage vendors' filers are usually commodity computers with a operating system such as Microsoft Windows Storage Server or tuned Linux. This is not the case with NetApp filers, which always use specialized hardware and proprietary Data ONTAP operating system, both designed specifically for storage-serving purposes.

All filers have battery backed NVRAM, which allows them to commit writes to stable storage quickly, without waiting on disk. Early filers used to connect to to external disk enclosures via SCSI, but the current ones use FC protocol. Those disk enclosures (shelves) support FC hard disk drives, as well as parallel ATA and serial ATA ones.

Two filers are often organized in a high-availability cluster via a private high speed link, either FC or InfiniBand. The newest models can also be organized into larger clusters with a single namespace when running the OnTap GX operating system.

1 Architecture
2 Current Limitations
3 Model history
4 References
5 See also

[edit] Architecture

Most NetApp filers are actually customized computers, with Intel or AMD processors and based on PCI. Each Filer has a proprietary NVRAM adapter to log all writes for performance and to play the data log forward in the event of an unplanned shutdown. Two filers can be linked together as a cluster, although NetApp now uses the more correct term Active/Active. Data ONTAP implements a single proprietary file system called WAFL. When used for file storage, Data ONTAP acts as an NFS server and/or a CIFS server, serving files to both Unix-like systems and Microsoft Windows systems from the same file systems. This makes it possible for Unix and Windows to share files by the use of three qtree security styles: mixed, ntfs, and unix. Qtree stands for quota trees and allows for data segregation and management inside of volumes. Qtrees with the UNIX security style will preserve the standard UNIX permission bits, the NTFS security style will preserve NT ACLs found in the Windows environment, and the mixed security allow both to be used interchangeably.

Each filer model comes with a set configuration of processor, RAM and NVRAM, which cannot be expanded once purchased. With the exception of the FAS200 series and the FAS2020, the NetApp filers have at least one PCI based slot available additional network, tape and/or disk connections.

NetApp supports either SATA, Fibre Channel, or SAS disk drives, which are grouped into RAID (Redundant Array of Inexpensive Disks or Redundant Array of Independent Disks) Groups of up to 28 (26 data disks plus 2 parity disks). Multiple RAID Groups form an aggregate and then flexible volumes are created within the aggregate to actually store data that can be accessed by users. An alternative is "Traditional volumes" where one or more RAID groups form a single static volume. The advantage of flexible volumes is that many of them can be created on a single aggregate and resized at any time. Smaller volumes can then share all of the spindles available to the underlying aggregate. Traditional volumes and aggregates can only be expanded, never contracted. However, Traditional Volumes can handle slightly higher I/O throughput than Flexible Volumes as they do not have to go through an additional viritualisation layer to talk to the underlying disk.

WAFL is a robust versioning file system and as such provides snapshots, which allows end-users to see earlier versions of files in the file system. Snapshots appear in a hidden directory ~snapshot for Windows (CIFS) or .snapshot for Unix (NFS). Up to 255 snapshots can be made of any Traditional or Flexible volume. Snapshots are read only although Data ONTAP 7 provides the ability to make snapshots writable as "FlexClones".

Snapshots are implemented by tracking changes to disk blocks between snapshots, and can be created in seconds because Data ONTAP only needs to take a copy of the root inode in the filesystem. This is different from the snapshots provided by some other storage vendors in which every block of storage has to be copied, which can take many hours.

Snapshots are the basis for NetApp disk replication technology SnapMirror, which effectively replicates snapshots between two NetApp filers. Later versions of Data ONTAP introduced cascading replication, where one volume could replicate to another and then another etc. NetApp also offers a backup product based around replicating and storing snapshots, called SnapVault. Open Systems SnapVault allows Windows and UNIX hosts to backup data to a NetApp filer and store any filesystem changes in snapshots.

Data ONTAP also implements an option called "SyncMirror" where all the raid groups within an aggregate or traditional volume can duplicated to another set of hard disks, typically at another site via a Fibre Channel link. NetApp provides a "MetroCluster" option, that uses "SyncMirror" to provide a geo-cluster or active/active cluster between two sites over 100km apart.

Other product options include "SnapLock" which implements a WORM functionality on magnetic disks instead of optical media, so that data cannot be deleted (even by the administrator) until it meets certain retention criteria.

NetApp also offers products for taking application consistent snapshots of Microsoft Exchange, Oracle, Microsoft SQL Server, Microsoft Sharepoint, VMware ESX Server, and SAP data. These products are part of the SnapManager suite.

[edit] Current Limitations

Currently individual aggregates sizes are limited to a maximum of 15TB, this has two impacts:-
1. With the availability of larger drives (such as 1TB), the number of spindles (ie physical disk) within an aggregate becomes smaller and as a result I/O cannot be spread over as many disks, reducing performance.
2. Each Aggregate incurs a storage capacity overhead of approximately 11%. On systems with many aggregates this can add up to a substantial overall capacity loss that isn't available to the end-user.

[edit] Model history

This list is incomplete and omits some early models and most current models. Info taken from spec.org

Model	Status	Released	CPU	Main Memory	NVRAM	RAW Capacity
FASServer	Discontinued	Jan 1995	50 MHz Intel i486	256 MB	4 MB	? TB
F330	Discontinued	Sept 1995	90 MHz Intel Pentium	256 MB	8 MB	? TB
F220	Discontinued	Feb 1996	75 MHz Intel Pentium	256 MB	8 MB	? TB
F540	Discontinued	June 1996	275 MHz DEC Alpha 21064	256 MB	8 MB	? TB
F210	Discontinued	May 1997	75 MHz Intel Pentium	256 MB	8 MB	? TB
F230	Discontinued	May 1997	90 MHz Intel Pentium	256 MB	8 MB	? TB
F520	Discontinued	May 1997	275 MHz DEC Alpha 21064	256 MB	8 MB	? TB
F630	Discontinued	June 1997	500 MHz DEC Alpha 21164A	512 MB	32 MB	? TB
F720	Discontinued	Aug 1998	400 MHz DEC Alpha 21164A	256 MB	8 MB	? TB
F740	Discontinued	Aug 1998	400 MHz DEC Alpha 21164A	512 MB	32 MB	? TB
F760	Discontinued	Aug 1998	600 MHz DEC Alpha 21164A	1 GB	32 MB	? TB
F840	Discontinued	Aug/Dec? 2000	733 MHz Intel P3 Coppermine	3 GB	128 MB	? TB
F820	Discontinued	Dec 2000	733 MHz Intel P3 Coppermine	1 GB	128 MB	? TB
F880	Discontinued	July 2001	Dual 733 MHz Intel P3 Coppermine	3 GB	128 MB	? TB
F810	Discontinued	Dec 2001	733 MHz Intel P3 Coppermine	512 MB	128 MB	? TB
F825	Discontinued	Aug 2002	733 MHz Intel P3 Coppermine	1 GB	128 MB	? TB
FAS940	Discontinued	Aug 2002	1.8 GHz Intel P4 Xeon	3 GB	256 MB	? TB
FAS960	Discontinued	Aug 2002	Dual 2.2 GHz Intel P4 Xeon	6 GB	256 MB	? TB
FAS980	Discontinued	Jan 2004	Dual 2.8 GHz Intel P4 Xeon MP 2 MB L3	8 GB	512 MB	? TB
FAS920	Discontinued	May 2004	2.0 GHz Intel P4 Xeon	2 GB	256 MB	6 TB
FAS6040		Dec 2007	?	16 GB	512 MB	840 TB
FAS6080		Dec 2007	?	64 GB	4 GB	1,176 TB
Model	Status	Released	CPU	Main Memory	NVRAM	RAW Capacity