John, apologies for the tardy reply.
I first became aware of the details of Selsius' story back in February in a BCR VoIP evaluation article. From that point, I've examined their wares from a distance, and although they are intriguing, they are not the same cut, or in the same class, as a lot of the other vendors commonly thought of as VoIP gateway or appliance companies. They make, well, IP PBXs, and ancillary networking elements and systems, and they mix some legacy protocols together with the new IP stuff, in ways that are still not fully hitting home intuitively. Easy enough to see what they're doing, but hard to envision the user marketplace adopting it now in large numbers, without some of them taking a few arrows first. But you know what they say about no pain, right?
While some aspects of their approach appear counterintuitive, that's only because I'm thick. Upon closer examination, I see the benefits that are possible in their architecture, but then I think of all of those legacy voice administration folks who still control the purse strings. I have to wonder about that, too, from a market viability standpoint. "Those people" are just as thick... no, make that they are much thicker than I am, and for that reason there will be considerable resistance to such a radical change in the "main" voice engine at this point of the convergence curve, for reasons other than those which stem from an intrinsic network ingenuity standpoint.
CSCO's acquisition announcement led me to immediately wonder how they would integrate this platform (which has ISDN capabilities and a non-IOS embedded system architecture) with its own. I don't know, frankly. The PR states that CSCO will use Selsius' platform in their fourth phase of their overall five-phase voice integration strategy.
My main point of concern here has nothing, or very little, to do with the feasibility of the architecture, because it makes sense from a purely technical standpoint. Rather it has to do with the fact that while IP PBXs are now viable, many firms will not take the risk of putting "all of their eggs" in one [what they feel to be still an unproven] basket. These same managers would, however, entertain leveraging a percentage or possibly all, at some point, of their long distance traffic over VoIP. There's a world of difference between these two risk assessments, and how far managers will stretch their necks out, IMHO. Maybe I'll feel differently about this in six months to a year, in fact I'm certain that I will. But not now.
If I might make one observation that is somewhat negative, however, the Selsius WAN strategy doesn't (or at least it didn't) take advantage of compression in order to offset high line costs. I found this strange, as the BCR article points out:
"...while Selsius' voice quality is fairly good, the system gobbles up a lot of bandwidth...(delete) .....At 100 kbps per voice connection, it would be expensive to link two or more sites across the wide area."
This will change, undoubtedly, if it hasn't already, but I found it to be a self-defeating trait at first reading, in the light of other vendors' products which yield hugh economies in this space. But like I said, it's not a gateway, it's an IP PBX.
For the most part, it seems to me that the entirety of their IP handling is contained within the confines of the LAN, and does not extend out onto the WAN. Someone correct me if I am mistaken about this, please.
The article that I referenced above can be found at:
bcr.com
Be sure to read the sidebars which appear as links under the Title.
I've clipped the passage from that article that speaks about the Selsius product line. It's printed below.
Regards, Frank Coluccio
------------------------------clipping begins:
A newcomer to the voice-over-IP gateway market, Selsius has taken a different approach than the other vendors tested. As the name implies, the Selsius LAN PBX is designed to functionally replace a PBX. It consists of one or more Access Gateways--prepackaged nodes equipped with LAN and ISDN PRI interfaces--and Call Manager server software, which runs on a (user-supplied) Windows NT server. Dozens of gateways can be driven and managed by a single Call Manager station, making the Selsius package fairly scalable.
Callers on the IP data network--an intranet, not the Internet--interact with the Call Manager and the gateways using Selsius' own $495 phone set (a fairly full-featured phone set; it's an IP node with its own Ethernet connection), or using special client software that loads and runs on a NetMeeting PC client. Straight NetMeeting clients are also supported, whether placing or receiving calls, via special proxy software in the Call Manager Server.
The Access Gateway also permits inbound and outbound calls between the IP network-attached users and ISDN-equipped PBX stations or ISDN-reachable PSTN stations. Regular T1 support is expected to ship in April, which should make regular PBX and PSTN stations reachable as well.
Call Manager is the key component in the Selsius package. It requires NT's Web server (version 3.0 of Microsoft's Internet Information Server, or IIS), because the Selsius system is managed via a Web browser that accesses the NT Web server. The "gateway" nodes are also configured and managed via the Call Manager's Web server.
The different components and platforms make things seem complicated and, with the added complexity of ISDN PRI, they are. Still, the vendor offers onsite installation and configuration, starting at $1,200 for a day on site, and that's definitely something we would recommend.
All the pieces work together, although, as with any product that's just out (shipping since late last year), there are rough spots. We were a little disappointed with the Web browser-based management interface, for example, which had limited ability to monitor status and activity in real time.
We also discovered that, while Selsius' voice quality is fairly good, the system gobbles up a lot of bandwidth: A single voice conversation can consume up to 100 kbps. The vendor says it is planning to roll out a more efficient vocoding method than the G.711 algorithm used in the product we tested, which should also help reduce the very high latency (220 ms of delay).
For now, though, we think latency and the high bandwidth requirement limit the Selsius LAN PBX to implementation on a single LAN, where users need connectivity to PBX and remote ISDN stations. At 100 kbps per voice connection, it would be expensive to link two or more sites across the wide area.
On the other hand, the LAN PBX is very fully featured and affordably priced. At $650 per port (which includes all necessary software, but not any of the phone sets or the NT platform) this is the least expensive of the four gateway products we tested. Accordingly, the LAN PBX package earned the "Best Value"--in terms of features and functions correlated to per-port price--award for this test series.
---------------clipping ends |
