enterprise buyers guides since 1991
StorageSearch.com interviewed - Chris Short at
PyX Tech (June 30,
the SSD buyers guide
After SSDs... What Next?
A Storage Architecture Guide - classic article
Overview of the Notebook SSD Market
the Problem with Write IOPS - in flash SSDs
Clarifying SSD Pricing - where does all the money go?
iSCSI is a child
of the 2000-2003 IT recession. Many
analysts predict iSCSI
will eventually replace the decade old technology of
Fibre-channel SANs. But
despite the hype and huge reader interest, I got the impression that products
were barely trickling out of most iSCSI factories and the projected flood
could still be "sometime in the future". |
In June 2004 that suspicion was confirmed when iSCSI startup company Xiran closed down (because their investors couldn't see any realistic prospect of making money from customers in the market or a sale of the company). But optimism about iSCSI was still aplenty too... in June 2004, iSCSI company EqualLogic announced it had got $20 million in a 3rd round of funding. So when I approached this interview with iSCSI technology company PyX Technologies I took a hard line business approach to find out what's really going on. And I'm pleased to say I got some down to earth revealing answers too.
has had a dedicated iSCSI index since March 2001 and there has been a lot of
interest in this by our readers in the last 18 months or so. But the market has
been slow to take off and is still quite small. For example, in April 2004,
Network Appliance (which claims to have nearly 50% of the iSCSI market)
revealed that it had only sold iSCSI storage for use in 100 end user systems.
And we know that some iSCSI companies have only sold a handful of products.
Doesn't that make it difficult for a dedicated iSCSI company to fund itself
during this ramp up period? How are you coping?
A. So this is a multi-layered question that addresses both the state of the market and the management process at PyX. Let's start with the management process as we believe it is a key factor in why we are experiencing success in light of so many failures and why we are getting significant traction in a difficult market segment.
When our president and CTO, Andre Hedrick, started the company three years ago, he recognized both the market demand for iSCSI and the reality that enterprise budgets were non- existent. This meant it was a perfect time to start a company that would need several years of R&D before products would be available as this timeframe would coincide with a significant retraction of corporate spending on all things IT. This also meant that many of the companies which were forced into chapter 7 or 11 would be liquidating their nearly new gear at rock bottom prices. PyX was able to take advantage of the downturn in two significant ways.
First, PyX became a significant net buyer of barely used enterprise hardware from EBAY at pennies on the dollar - a now recognized and highly promoted method of building a new company. This fact is documented in the July 2004 issue of Entrepreneur Magazine. Simultaneously, PyX became a true garage startup with our corporate headquarters dislocating the Saab and the minivan to the driveway. By being frugal with overhead, PyX was able to complete the necessary three years of R&D, and the alpha and beta trials without burning significant capital. This all occurred during a period when nothing but the most tried, true and essential technologies were selling. PyX management recognized this was not the optimal time to introduce a paradigm changing technology.
Second, PyX solutions are 100% software in a market space that is traditionally hardware-centric. Because we do not need to buy, build or hold inventory, our cost of doing business is very low. This also allows PyX to work with any OEM that is interested in enabling their hardware with block level storage transport. This opens market segments to PyX which are not thought of as traditional storage markets and allows PyX to leverage the distribution networks of the OEMs to whom we license. Because we are privately held and intensely focused building a dynamic and strong business rather than reporting quarterly results to investors, we can take a long term perspective on both the iSCSI market and the growth of our organization. Said another way, we can be very patient with a nascent market. iSCSI will transform the storage market ethos, and we will scale to the demand as it comes.
As with almost any emerging technology, there is a lot of FUD in the market. Markets are skeptical and rightfully so. There are some large players who may not view the deployment of iSCSI as being aligned with their interests at this stage of the game. (What, no product?) There have also been several companies which have tainted the market perception somewhat by producing and distributing poor product that was not standards based.
Adding to the FUD was the low-ball buyout of a formerly highly visible player in the space and the rumors of pending flame-outs by some of the remaining vendors. It's important to keep in mind that the VC folks have specific agendas when they fund an organization and for all the sincerity about helping entrepreneurs achieve their vision, at the end of the quarter, most VCs still have to report to their investors. It's apparent that the delay in the market development was not modeled into some investors. calculations and they have been re-evaluating and dumping holdings.
If you use the VoIP market as an analogue, we know this market is going to be huge and it will be used by almost everyone as it is accepted; The question everyone is asking themselves is whether 2004 is to iSCSI as 1998, 2000 or 2002 was to VoIP. We have fully standards compliant products which are shipping today, so we are not worrying so much about where we are in terms of mass market adoption as we do about getting the message out about what we can deliver and licensing it to savvy OEM product managers.
Q. What do you see as the "killer app" which will transform iSCSI into a multi-billion dollar market? And how do you see that timescale unfolding?
A. The obvious answer to us is the disaster recovery market - in all its various shades. If you take a look at the state of disaster recovery methodology today, it is, quite frankly... a disaster! It's also a matter of patriotism with us. Many of us in the organization were not only enraged by the events of 9/11/2001, but we also saw how vulnerable many systems were to what should have been a less significant event from an IT perspective. Our country's ability to handle failure and disruption was not as robust as what we knew it could be. One of our corporate goals is to do what we can to make sure that the data necessary to run the economy of United States is never again at as great a risk from an attack from any source.
Most everyone agrees that writing and recovering to and from disk is the preferred methodology for disaster recovery, but with current technology, that is an expensive proposition. iSCSI allows for simultaneous reads and writes to multiple disks, in multiple locations, using commodity hardware over TCP/IP. As this technology matures, we'll see the disaster recovery market becoming an asset to the IT community as opposed to the necessary evil it is today. This will be happening in the next few years as the large OEMs start producing hardware capable of block level storage transport over TCP/IP. In five years, this area will look significantly different than it does today.
Bear in mind however, for all the promise of this single segment of the market, it is still only one of many markets that will be dramatically improved by this technology. We really are on the verge of SAN over IP technology making all information available at line speed anywhere in the world, anytime.
|Q. How will users benefit
most from PyX Technologies' products?
A. We believe that consumers will not only benefit from our products, but we are diligently working to increase awareness of what iSCSI really is capable of doing. This means educating the end user that there is more to iSCSI than what has been available or widely discussed in the market and the trade journals to date. Most of the emphasis surrounding iSCSI has been on the benefits of cost, with the implied negative that quality of data integrity must be sacrificed. This has happened because the early products in the space were not fully implementing the protocol and because there was only the technical expertise to produce Error Recovery Level Zero (ERL 0) products. This is analogous to using a model T Ford as being representative of the 2004 auto industry. It's still a car, but.
To our knowledge, we are the only company in the world currently licensing the complete implementation of the IETF RFC 3347. This means we are licensing products that have ERL 0, 1 and 2 capabilities and allow data transport with the same integrity as Fibre Channel at roughly 10- 15% of the cost. The immediate benefit to consumers therefore is the democratization of enterprise storage solutions.
The extended benefit is that entirely new technologies in the entertainment, healthcare and security markets are being developed using iSCSI to deliver previously only dreamed of functionality to a globally distributed network of end users over TCP/IP. This means we are using a storage transport to change markets in areas that are not traditionally considered purvey for a storage solutions provider.
Q. Is your brand going to be visible or is it going to be buried inside white box products from 3rd parties?
A. We are working to develop brand awareness of PyX Technologies. However our customers are OEMs and component manufacturers so by definition we are going to be embedded in the hardware of another vendor at even the most basic level. Ideally we will have the market presence to support a "PyX Powered" identity on every product we enable, but that is a topic for the contract attorneys to haggle over. We are focused on providing the best products to the market and licensing to the market leaders in the spaces in which we compete. If we are able to achieve that goal, then the individuals who need to know who we are will be our best customers.
Q. I see that your software supports a lot of platforms, Intel Architecture, SPARC, Windows, Linux etc. Are you seeing any difference in take up for your products between these markets? Can you tell us which looks most promising in the short term and why?
A. This really depends on whether you are looking at the target or the initiator market. Because the initiator is usually associated with a client, we see more demand for Windows initiators due to the ubiquity of Microsoft's O/S. The market is most interested in targets however and in that area the demand for Linux is very high. Sun has a strong presence in the storage space so there is also significant demand for Solaris. At this point, the value proposition still has to be validated for most buyers, so it is a little early to gauge where the big volume demand will finally originate, but the way we see it in June 2004 Linux has a clear lead.
They show the physical disks
too far above the protocol to
allow the iSCSI protocol to
multi-path or aggregate.
|Q. STORAGEsearch has heard
a lot of hype in the last 18 months from companies which have developed iSCSI
accelerator cards. How should users view these products? As proprietary or part
of an open standard? Can users mix and match iSCSI accelerators in the type of
storage architecture which your products enable?
A. Well, we are working with companies that are developing accelerator products. Our goal is to be embedded into these products at the foundry or the assembly line. Of course that is not always possible so we engage with customers using accelerator products and discuss their goals and hardware requirements. We are big fans of open standards. Not only does it make our lives easier, we believe it is better for the economy as a whole.
The interaction with other iSCSI accelerator cards can be problematic however as many RAID controllers and iSCSI engines supply the iSCSI stack inside the card itself. They also show the physical disks too far above the protocol to allow the iSCSI protocol to multi-path or aggregate. The iSCSI protocol is therefore undermined and will not operate. That being said, we work well with most TOE cards and straight NAC cards and have products on the market with partner companies such as SBE.
So the quick answer is yes to TOE and NAC cards and usually no to iSCSI cards because the end user can't load an alternate engine on a pre-configured card.
Q. What are the main risk factors you see for users who start implementing iSCSI today?
A. The biggest risk we see is that buyers develop the wrong expectations for the technology and accept sub-standard products as being representative of the entire class of iSCSI enabled devices. There are also a lot of products that simply don't work well or as well as they should when they do work. The expectations being set are well below what is possible and it is slowing the acceptance of the technology.
Further, we really believe the technology is being mis-marketed. Almost all product available today is ERL 0 only and has extremely limited functionality. There is little awareness that iSCSI has different levels for fault tolerance within the protocol or that not all iSCSI-enabled products are fully compliant with the protocol as written. There is little to no mention of standards, it is being sold as technology with limited functionality and there are some performance claims being floated in the market that are mathematically impossible. This is not the way to develop market demand for serious products.
We are seeing some rather seedy marketing of the technology as being a "cheap" solution or "Poor Man's Fibre Channel". While the solution may be less expensive than Fibre, in our opinion the only things that should be referred to as cheap are thrills and liquor. Nonetheless, that has not prevented some vendors of low-end iSCSI enabled products from putting poor out quality product and selling it like they're at the Sunday flea market.
That being said, it sounds like we're the bitter curmudgeon of week. Actually, we're just concerned that the early adopters don't underestimate the technology based on what has been available to date or be deterred from adoption of iSCSI because they have a bad experience with an early implementation that was not correctly developed, engineered or deployed.
|Q. STORAGEsearch has seen a
lot of storage companies whose main business plan is blatantly to be acquired
by EMC, Veritas, Sun or NetApp. Those companies just don't do any marketing.
Acquisition seems to be where a lot of good storage technology companies end
up. Do you have a business plan which includes marketing long term as an
independent technology company? How do you see that progressing during the next
few years with the split between direct sales, indirect channels and oems?
A. Gee, that's assuming any company would actually want us working for it - clearly you haven't spoken with us before morning coffee! Actually, our business horizon is ten years. We believe the market will develop rapidly well before then, but our benchmarks are set with a decade-long horizon.
Most of us in the organization have neither the patience nor the desire to go back to corporate life. We have worked very hard to remain independent and have no desire, nor intention, of being acquired at this stage. We are businessmen however, and most everything is available for the correct price. Specifically though, our intent is to remain independent and work our business model.
Our marketing has been limited to date, but this has been mostly a function of the speed of the growth of the organization and the state of the technology. We will be doing more marketing as we expand and intend to be "The Name" in the iSCSI OEM market space. To that end, we have hired a PR firm to help raise our visibility and to help us raise the market awareness of iSCSI and the capabilities of the technology.
Our bread and butter market is product developers and product managers. We plan to enable hardware that manufacturers design, build and market. Whether that is a chip, a box, a blade, a camera, a line card, a radio or a laser, we want to work with the design and marketing team to enable it to be better. We do not intend to develop a huge channel on our own. We do intend to enable and support our customers with huge channels to deliver better products to their customers. ...PyX Technologies profile, iSCSI
|Marketing Views||STORAGEsearch||SPARC Product Directory||ACSL - the publisher|