Cheaper and easier to use than Fibre Channel, iSCSIstorage system is considered by many to be the next great storage technology for small to medium-sized businesses (SMBs), which are already rapidly deploying iSCSI-based storage, according to a Storage August 2006 Special Supplement.
Here we've compiled 10 iSCSI SAN storage questions addressed in that supplement, which you may find useful as a VAR or systems integrator handling SMB storage. These questions and answers will help you sell the benefits of iSCSI SAN storage and develop service offerings around the technology. For more in-depth iSCSI SAN advice, click here for the full supplement.
1. Why should SMBs look at iSCSI storage systems vs. Fibre Channel (FC) array SANs?
- iSCSI SANs are less expensive than FC SANs.
- An iSCSI SAN is based on the familiar and ubiquitous TCP/IP protocol.
- iSCSI SANs are less complex and easier to manage than FC SANs.
2. What will iSCSI SAN storage replace?
iSCSI SANs are replacing direct-attached storage (DAS), which is hard to scale and susceptible to failure. Shops that are experiencing accelerated data growth will want to expand capacity and improve backup and recovery procedures, which is much easier to accomplish in an iSCSI environment.
3. How can iSCSI save you and your customers' money?
iSCSI uses TCP/IP instead of a proprietary storage protocol such as FC. As a result, iSCSI SANs can be built with commodity network components, eliminating the need for costly FC host bus adapters (HBAs), which are typically priced between $800 and $1,200, and FC switches.
4. How are iSCSI storage systems being deployed?
The vast majority of iSCSI deployments today are on Windows, followed by Linux. One of the reasons iSCSI is predominantly seen in Windows environments is Microsoft Corp.'s firm support of iSCSI since its ratification as a standard in 2003. With the 2.0 release of its iSCSI Software Initiator, Microsoft was the first OS vendor to offer a full-fledged initiator that includes high-end features like multipathing I/O (MPIO), which enables the same level of reliability as FC.
5. How is iSCSI being used in SMBs vs. large enterprises?
While SMBs use iSCSI SANs as their primary storage, large enterprises use iSCSI to complement their FC array storage for less mission-critical departmental apps, remote offices and data protection. Typical applications for iSCSI are messaging apps like Microsoft Exchange, databases, Web apps, file serving and disk-to-disk backup.
6. What iSCSI weaknesses should you consider when developing iSCSI services offerings?
Storage management is one of the weaker aspects of iSCSI. Storage management vendors are just starting to add iSCSI support to their suites. While FC switches are well supported by all major storage management apps, only some aspects of Ethernet switches, like availability, are addressed by storage management suites.
Security is another controversial aspect: Although FC proponents are quick to point out security concerns with iSCSI, in reality, iSCSI is an inherently more secure protocol than Fibre Channel.
With the reliability and feature gap closed, the slower performance of iSCSI is now the primary technical argument FC advocates use for not considering iSCSI for enterprise-level applications. But iSCSI speeds are beginning to catch up with FC, and the two protocols are on a leap-frog path with each one boasting new product releases that bests the other, but not for very long. However, the FC protocol has a slightly lower latency than TCP/IP.
7. What iSCSI deployment challenges should consider when developing iSCSI services offerings?
iSCSI can't boot from iSCSI-attached storage through a regular network interface card (NIC). The boot process is performed by the system BIOS and with the iSCSI software initiator running at the OS level, iSCSI storage is out of reach until the OS has booted. The only practical solution in the past was adding an iSCSI host bus adapter from companies like QLogic Corp., which provides the interrupt 13 (INT13) extensions required for booting, as well as an on-board iSCSI initiator.
This changed in April when Microsoft Corp. announced support for software-based SAN boot of Windows using the Microsoft iSCSI Software Initiator and standard NICs through a technology co-developed by Microsoft and IBM Corp. Software-based SAN boot requires a NIC firmware upgrade or an upgrade of the System Options ROM of blade servers.
8. Who are the iSCSI players?
iSCSI target vendors can be grouped into three categories: pure iSCSI players, traditional storage vendors adding iSCSI support to their existing arrays and iSCSI gateway vendors.
Pure iSCSI players: Pioneers include EqualLogic, Intransa, LeftHand Networks, StoneFly and Sanrad Inc., as well as younger startup companies such as Nimbus Data Systems Inc. Most offerings in this group are running iSCSI target software on a server -- in most cases concealed in an appliance form factor -- with attached storage and some flavor of a general-purpose OS. Vendors in this category have added storage features previously available only in FC SANs, including snapshotting, remote replication, load balancing and virtualization, which enables them to compete head to head with FC vendors.
Traditional storage array vendors: EMC Corp., Hewlett-Packard (HP) Co., Hitachi Data Systems and Network Appliance (NetApp) Inc. have added iSCSI support to their arrays to more effectively support their large customer bases. Conspicuously absent from the list is IBM Corp., which hasn't added iSCSI support to its high-end arrays. With the exception of the low-end TotalStorage DS300, IBM depends on its partnership with NetApp to fulfill higher end iSCSI customer requests.
Gateway vendors: Similar to large array vendors, FC switch vendors Brocade Communications Systems Inc., Cisco Systems Inc. and McData Corp. have either added iSCSI support to some of their FC switches or are offering dedicated iSCSI gateways to enable enterprise customers to attach iSCSI servers to existing FC storage.
9. Why iSCSI, why now?
One could argue that the popularity of NAS was the main reason for the ever-increasing interest in iSCSI. NAS dependency on higher level file-system protocols (CIFS/NFS) makes NAS less suitable for block-based transactional applications like databases; as a result, NAS vendors are adding iSCSI support to their offerings. Because NAS and iSCSI are both based on TCP/IP, iSCSI is a more natural SAN supplement than FC in NAS environments.
Furthermore, the number of innovations and companies developing for the TCP/IP and iSCSI space today greatly surpasses the number of FC developments. Besides 10GigE, network packet processors--from companies like Broadcom Corp. and Cavium Networks--for packet inspection, prioritization, encryption, compression and TCP optimization are the building blocks for next-generation TCP/IP and iSCSI products.
10. Will iSCSI push FC out of the data center?
There's no doubt iSCSI will dominate the SMB storage space and will be the preferred storage option for workgroup servers and, eventually, corporate desktops. What's still unclear is if iSCSI can push FC out of the enterprise data center. With FC's large installed base, long-standing support for mission-critical apps and the availability of 8Gb/sec FC, it's the preferred technology for most large data center storage. For a transition to happen, it will take large storage vendors like EMC, HP and IBM to recommend iSCSI over FC, which is unlikely to happen in the near future.
Are you providing iSCSI solutions and services? Send us an email to share your observations and experiences.
Dig deeper on Network-Attached Storage (NAS)