re: Who Makes What: GPON Equipmentre: "Now, that is their discussion of their product." -- You are quoting only one of Novera's two main products and are avoiding making any mention of faster of the two. But that doesn't really matter at this time, as I explain below. I will say, however, that I do appreciate the additional insight you've brought to the discussion.
What we might otherwise be discussing here, IMO, is a migration path to Enterprise PONs (or FTTD/FTTE using pt-2-pt optical Ethernet too) and NOT jumping immediately into the highest achievable data rates during the first iteration in this space.
Any serious analysis of most business operations will reveal that most "information workers" in ordinary business offices who are attached to LANs that are properly tuned seldom require even a couple of tens of Mbps, some may peak to 100 Mbps, but only a relative few at this time require a full Gigabit pipe (although I'll grant you one is always nice to have). Similar analyses were performed decades ago for terminal-based networks in deciding whether sub-second or faster response times were sufficient, and what the extra costs associated, on average, would be for every hundredth of a second shaved.
Future dynamic bandwidth allocation (DBA) capabilities and programmable speeds in some PON designs will prove more than suitable for achieving acceptable economics over distributions of 32 or more end points. One can always scale back to 16 or fewer in those situations that warrant doing so for groups where power users dominate, and still be ahead of the game when comparing the total number of optical ports required in the overall scheme.
One goal that needs to be satisfied in many established enterprise settins is peaceful coexistence with prevailing environments. One aspect of this means refraining from going in with an all-optical mandate, lest all terminating hardware (e.g., wall outlets) be deemed inaccessible by laptops and other appliances that have not been specifically designated as FTTD endpoints.
This may require allowing "all" end points to continue using Modular 8-pin (RJ-like) TIA/EIA UTP patch cords and termination hardware, thus allowing users to continue using their PC NIC cards as well.
This can be achieved by situating the ONT or equivalent within the wallplate enclosure, or behind it, or through the use of a dongle (although the last of these may be deemed unacceptable by many IT honchos). And of course, some early adopters will opt for an all-optical approach instead.
"Has anybody considered the operational limitation in GPON due to the fact that if one ONT fails all the 32 other ONT are out of service due to lack og syncronization. This happens in practice."
If that is happening in your network, you need a new GPON vendor.
Novera tends to put systems on their website years before availability. So, I ignored their brand new system (the one with no deployments) until they have deployments to show that it is real.
There is a small potential for an ONT to bring down a PON if the optical transmitter on the ONT fails in the on position. Please note that FiOS does not have this happen very often (perhaps LR could ask Verizon if this has EVER happened). The same would be true for EPON by the way.
re: Who Makes What: GPON Equipment... of mine, before I return to the silica mines, can be found in the second entry below:
In case anyone reading this hasn't yet read Network Worlds excellent write-up of Verizon's (and partner SAIC's) proposed entry into the Enterprise PON space, I strongly commend it:
re: Who Makes What: GPON EquipmentQuestion: Would the one "that has no deployments" be the same product that ADC has licensed? I think it may be, although, as IGÇÖve already stated, itGÇÖs a bit too early at this time to really matter. In any case, Novera is not the only manufacturer of WDM-PON designs, nor does it have a lock on its direction.
I believe that a reasonable trajectory for first-pass xPON penetrations in the enterprise, beyond those that have already been deployed by some MSOs to SMBs, and VerizonGÇÖs somewhat limited extension of its FiOS offering, likewise to a select few government installations and SO-HO, begins sometime in the late '09 to early GÇÿ10 time frame, with WDM-PONs coinciding on the trailing edge, although I see the latter first filling a different set of needs than the former.
Thus far we've merely touched upon, and very superficially, at best, some of the technical aspects of PONs of various types as a potential fit in the enterprise. GÇ£ActiveGÇ¥ or "switched" (i.e., point-to-point optical Ethernet) will very likely be seen as far more acceptable by most enterprise managers in the early going, since the wiring templates to support active designs are already included in GÇ£industry best practices", despite however limiting those may be, as defined in the EIA/TIA standards governing commercial buildings wiring systems. While those standards currently accommodate FTTD and FTTE(nclosures), they do not at the present time, however, conform to the designs that meet the needs for the best that PONs have to offer.
The main barriers to the adoption of both PONs and GÇ£meaningfulGÇ¥ switched optical solutions, alike, where the distance insensitivity of fiber can be best leveraged in the enterprise, will not center on hardware or software capabilities exclusively, but will also be impaired for some time by cultural issues - and I should add some other valid reasons as well -- but also, I should further add, for some reasons that are not so valid.
Most of the technical objections to the use of fiber in the enterprise are either remedied quite easily, or they are anachronisms that no longer fit in an all- or near-all optical regime. The most perplexing issues, however, and as usual, can be found residing in wetware, and those are the most difficult of all to fix.
re: Who Makes What: GPON EquipmentIt happens with 3 GPON vendors .... It is intrinsic in the GPON protocol... By the way there is also the possibility that if a there is a failure with the laser on all return channels are affected...
re: Who Makes What: GPON Equipment(Changing the subject line to break this one out of the bunch.)
Verizon did indeed have a problem with what they called "rogue ONTs". Those were ONTs that would broadcast upstream at the wrong time. Say the rogue is in timeslot 1 and broadcasts in timeslot 2. Since the ONT that's supposed to be in timeslot 2 is also broadcasting, the OLT gets garbage. If the OLT gets garbage enough times, it tells the ONT to shut down. Unfortunately, it tells ONT #2 to shut down, not ONT #1 (since the garbage is in timeslot 2). Eventually, one rogue can shut down an entire PON.
Verizon had this problem with their early BPON deployments and forced the issue onto all of their vendors. It was a big part of the tri-BOC GPON RFP lo these many years ago.
Any vendor who has experience and contacts within Verizon (you know who they are ... and aren't) implemented a fix before GPON was deployed. The fix essentially puts in a monitoring circuit on the ONT transmitter. If the monitoring circuit finds a signal outside of the correct timeslot, the ONT is shut off. There are other "fixes" as well to account for massive software failure and lightning damage.
I'll echo the earlier comment. If you're seeing this problem in your network, change vendors.
The timing issue is at the OLT and related to the recovery time of the signal of the individual grant. This has little to nothing to do with the stability of the ranging process.
The Luminent optics had many quality issues at the start of the BPON deployment. As far as I know, there have been 0 rogue ONTs outside of optics failures.
re: Who Makes What: GPON EquipmentGPON has some critical timing parameters due to it's high speed, but the ranging shoudl handle this no problem. Changing the vendors who designed the controller function is liely the best approach. Getting the system to function properly is system tweaking, not optics level stuff (except of course to change out any bad optics or bad laser or whatever fabius was referring too).
--
You are quoting only one of Novera's two main products and are avoiding making any mention of faster of the two. But that doesn't really matter at this time, as I explain below. I will say, however, that I do appreciate the additional insight you've brought to the discussion.
What we might otherwise be discussing here, IMO, is a migration path to Enterprise PONs (or FTTD/FTTE using pt-2-pt optical Ethernet too) and NOT jumping immediately into the highest achievable data rates during the first iteration in this space.
Any serious analysis of most business operations will reveal that most "information workers" in ordinary business offices who are attached to LANs that are properly tuned seldom require even a couple of tens of Mbps, some may peak to 100 Mbps, but only a relative few at this time require a full Gigabit pipe (although I'll grant you one is always nice to have). Similar analyses were performed decades ago for terminal-based networks in deciding whether sub-second or faster response times were sufficient, and what the extra costs associated, on average, would be for every hundredth of a second shaved.
Future dynamic bandwidth allocation (DBA) capabilities and programmable speeds in some PON designs will prove more than suitable for achieving acceptable economics over distributions of 32 or more end points. One can always scale back to 16 or fewer in those situations that warrant doing so for groups where power users dominate, and still be ahead of the game when comparing the total number of optical ports required in the overall scheme.
One goal that needs to be satisfied in many established enterprise settins is peaceful coexistence with prevailing environments. One aspect of this means refraining from going in with an all-optical mandate, lest all terminating hardware (e.g., wall outlets) be deemed inaccessible by laptops and other appliances that have not been specifically designated as FTTD endpoints.
This may require allowing "all" end points to continue using Modular 8-pin (RJ-like) TIA/EIA UTP patch cords and termination hardware, thus allowing users to continue using their PC NIC cards as well.
This can be achieved by situating the ONT or equivalent within the wallplate enclosure, or behind it, or through the use of a dongle (although the last of these may be deemed unacceptable by many IT honchos). And of course, some early adopters will opt for an all-optical approach instead.