NetConvergence's Simon Fok Sheds Light On iSCSI - Technology Information - Interview

Computer Technology Review, March, 2001 by Ilona Fass

iSCSI will most likely replace Fibre Channel

Well-known for his candid views on the connectivity community, particularly in regards to the future of Fibre Channel, Simon Fok, president and Chief Executive Officer of NetConvergence, has more than 15 years of experience in designing and producing high-speed networking switches. Fok has received numerous patents on bus/networking protocol switching concepts. He co-founded both GigaLabs, Inc and Systolic Technology Corp., and served as director of Input Output Systems Corp. He received his Ph.D. in Applied Mathematics from the University of California at Berkeley.

Founded in 1998, NetConvergence is an IP Storage company that designs and markets a full range of iSCSI products. With the development of their MATRIX software technology, NetConvergence hopes to enable customers to utilize their existing IP-networking infrastructures as a solid foundation for building high performance storage networks.

CTR: What do you think the future of Fibre Channel is?

Fok: As time goes on, IP storage will slowly gain acceptance and Fibre Channel deployment will slowly level off. Already, Fibre Channel's situation can be compared to what is happening to SCSI. What I mean by that is that, just as Fibre Channel replaced SCSI, iSCSI will most likely replace Fibre Channel. With that said, neither SCSI nor Fibre Channel will disappear any time soon.

CTR: Do you think Gigabit Ethernet will replace FC?

Fok: I do think that eventually Gigabit Ethernet will replace FC. However, this is not something that will happen any time soon. It will be at least four or five years before Fibre Channel starts leveling off in terms of deployment.

CTR: Where do you think Gigabit Ethernet will replace Fibre Channel?

Fok: Fibre Channel will come from medium to low end environments before it will percolate up to the enterprise level, which is where it would stop. That's the reason why it will most likely take some time. You've probably seen EMC's market evolving in IP storage deployment, maybe two years out.

CTR: How would you define iSCSI? Why do people need it?

Fok: The definition for iSCSI is currently being drafted by the IETF (Internet Engineering Task Force), and is being worked on by many companies. We need iSCSI because in storage it's getting to the point where connections will be over long distances. When you go long distances in the world today, you come across IP and TCP/IP. iSCSI is the proposal to have connectivity on top of TCP/IP. Not only could the server be connected to the client, but also to the storage application. This is something that Fibre Channel is not yet capable of because it is still limited to 10 km of distance. In terms of longer distances you need something different. That's what iSCSI is all about.

CTR: NetConvergence is a software firm. How can you meet performance requirements without having custom hardware?

Fok: Our software is cased in hardware by OEM partners like EMC, and that is where the performance is going to be derived from. There's a misconception that because we sell software, performance will be poor. But the fact is that "off the shelf" components are very fast and can meet requirements without hardware. People who think of off the shelf as being limited to Pentium are mistaken. Off the shelf is really pointing to whatever you can buy without having it specifically made. And the truth is that many processors that are not Pentium based have been proven to deliver wire speed performance that can be leveraged to get software to high speed. There are a lot of chips that can be leveraged to get wire speed performance without sacrificing flexibility.

The world is networking at a point where many things can be bought off the shelf. And the advantage of this is that it's programmable, which gives us the possibility to enable storage and networking folks to maintain their infrastructure without having to replace it to do IP storage. So, in other words, they already have the hardware and the knowledge to do IP. There is no reason to try and reinvent the whole thing. All they need is intellectual property or software from us, which they can incorporate into existing hardware. Of course, people keep saying that they want to make sure people buy their own special pieces of hardware. Rather than a custom ASIC solution that has little flexibility and scalability, this is what NetConvergence is about.

CTR: iSCSI is new technology. How will it meet interoperability challenges?

Fok: iSCSI is a technology that SNIA is currently involved with. SNIA (Storage Networking Industry Association) is a group made up of multiple vendors who are working together to create interoperable products before the standards can be finalized. In fact, SNIA is not limited to iSCSI; it's about other protocols that would potentially become standard. The path SNIA is following can be compared to that of Gigabit Ethernet, which also had a group working to create interoperable products. The advantage of having multiple vendors is that no one will get railroaded into using one vendor. In addition to participating in SNIA, we at NetConvergence are working independently with other vendors, such as Intel and others, to produce products. Of course, there are other companies like Cisco and IBM that will have products out soon. The vendors are simultaneously competing and cooperating to deliver a total solution to customers. Interoperability and cooperation are the key words. That's fantastic since it indicates a hea lthy industry. As a side note, this was not the case for Fibre Channel.