RE: [nznog] Acceptable levels of Latency, Jitter and Packet loss on national links
-----Original Message----- From: Joe Abley [mailto:jabley(a)isc.org] Sent: Monday, 27 September 2004 4:09 p.m. To: Craig Humphrey Cc: nznog(a)list.waikato.ac.nz Subject: Re: [nznog] Acceptable levels of Latency, Jitter and Packet loss on national links
On 26 Sep 2004, at 22:25, Craig Humphrey wrote:
Latency Jitter IP Packet Loss "Realtime" <50ms <12.5ms <0.1% "Interactive" <75ms <50ms <0.1% "Business" <200ms <100ms <1.0%
And yes, we're chasing references who are already using
this service.
The reason for my dissatisfaction... Based on crude testing
with ping,
we already get an average of 25ms latency over the 2Mbit/s fibre (during normal background load), and that's a round trip! So I'm not sure how the above is really an improvement...
What is the actual RTT over the ethernet service? Normal performance and guaranteed performance seem very unlikely to be similar (or even comparable within an order of magnitude).
Among other reasons this is because normal performance is seen when things are normal. A provider's performance guarantee is presumably going to have a more demanding specification than just "you'll see numbers like this most of the time". - Donald Neal [All opinions mine only.] Donald Neal |World Domination is Technical Specialist |proceeding according to Operations Engineering |plan. - Andrew Tridgell Integration & Services Division +----------------------- Alcatel NZ Ltd - Telecom's network operations manager ------------------------------------------------------------------------------ "This communication, including any attachments, is confidential. If you are not the intended recipient, you should not read it - please contact me immediately, destroy it, and do not copy or use any part of this communication or disclose anything about it. Thank you. Please note that this communication does not designate an information system for the purposes of the Electronic Transactions Act 2002." ------------------------------------------------------------------------------
On 27 Sep 2004, at 00:59, Donald Neal wrote:
Among other reasons this is because normal performance is seen when things are normal. A provider's performance guarantee is presumably going to have a more demanding specification than just "you'll see numbers like this most of the time".
It may actually be that the performance guarantee is a paper-dwelling construct, and has little or no form in the actual network. I have certainly come across providers in my past for whom performance guarantees were necessary in order to be able to bid for projects, but the cost of deploying infrastructure to support the guarantees was much higher than just giving a few refunds now and then. As such the guarantees didn't really exist outside the billing department. Joe
participants (2)
-
Donald Neal -
Joe Abley