You are right on the bandwidth alloted to each user is sufficient enough to service subscriber considering if the number of splits are still decent as recommended by the vendor. There is however more complications to just that consideration alone. IPTV delivery has to be taken with different approach because its multicasted with guaranteed QoS so if supposed the whole splicing ratio is satisfied.
- The internet virtual circuit(VC) return path rate available on your OLT that leads back to concentrators for all users.
- Laser power requirements on each subscriber end
- Individual account speed throttling (not done by the OLT itself which requires massive computation)
It is not safe to assume that a 1.25Gbps rate is sufficient to cater say 32 users boldly by calculating:
1250mbps/32 = 39mbps per user has lots to spare.
Vendors will tell you this general guide:
If your short term plan is 50mbps or nothing more than 64 splices is the best splice ratio to achieve good returns.
If your needs is 100mbps then 32 splices is the recommended figure to follow in the case of Time allowing ASTRO Beyond.
This is in reference to a OLT port set at 2.5Gbps rate. When problem arise, they will quietly adjust it down to half rate to solve the laser requirement issues.
Each OLT port is only capable of seeing the total downlink/uplink traffic it handles. The passive nature makes it treat all the other splits on the customers end all equal and has no capability to sense each of them differently.
I do not have much information on Malaysian ISP setup to discuss anything further but I know we are not alone when facing this drop in speed across FTTH users elsewhere especially those who implement speed caps similar to mainstream DSL speeds(10-20-30-50mbps)
Singapore ISPs do not suffer from this lag is because their speeds are in generally fast enough no to have their customers feel the slowdowns on peak times. How much slower does it get if a 100mbps or a 200mbps line drops by halve on peak hours? The average user will not be concerned by it.
On the VDSL2 users also being affected. I have little to add about it since I don't know the general setup to comment for their uplink ports for the VDSL2 switches installed inside buildings.Are they served with existing Unifi FTTH connection for its uplink ports? Does those links have additional splices to contend with?
I'm not sure what you mean by 'Each OLT port is only capable of seeing the total downlink/uplink traffic it handles. The passive nature makes it treat all the other splits on the customers end all equal and has no capability to sense each of them differently.'. This may be case at the physical level, however at the logical level each OLT port is capable of identifying every ONT that's connected to it .. including polling of individual laser TX bias, ranging information (via the optical module on the ONTs themselves) and service flows enabled for that OLT. Furthermore, the OLTs do handle individual user rate limiting but not at the account level.. more at the ethernet/VLAN level. As for throttling, some OLTs can do that but it's vendor specific.
I have yet to obtain a GPON FPGA for lower level frame analysis of the technology but I have written multiple pieces of network data mining & diagnostics software for Alcatel/Huawei/ZTE/Fiberhome operating directly with their OLTs and IPMSANs.
I can't really delve into specific numbers without breaking a few NDAs so all I can do is assure you that over here we do look at OLT-ONT layer statistics as well as further up the network.. and that the ports are not operating oversold + at a reduced throughput rate. The majority of our bandwidth contention does not occur at the access but further up at the access concentrators. Perhaps it's different in Singapore
Aug 12 2012, 09:44 PM
Quote
?
0.0380sec
0.29
7 queries
GZIP Disabled