> I am curious about SLA vs. message size. If Pulsar has a 5ms SLA is that going to be feasible if there are very large messages?
Of course it's not feasible :) The numbers published were related to 1KB entries. Of course there can be a huge number of different use cases with different trades off. Typically most users don't care about latency of bigger messages, so we never really tested it, but at least at lower transfer rate, the latency shouldn't be too bad even for 1MB messages. -- Matteo Merli <matteo.me...@gmail.com> On Mon, Jun 26, 2017 at 1:42 PM, Sahaya Andrews <sahaya.andr...@gmail.com> wrote: > As the message size grows, we won't be able to guarantee 5ms latency > since the underlying storage itself will take longer to sync the data. > > Andrews. > > On Mon, Jun 26, 2017 at 1:31 PM, Dave Fisher <dave2w...@comcast.net> > wrote: > > Hi - > > > > If you are negotiating the buffer size then you should also negotiate if > this is a compressed or uncompressed size. > > > > I am curious about SLA vs. message size. If Pulsar has a 5ms SLA is that > going to be feasible if there are very large messages? > > > > Regards, > > Dave > > > >> On Jun 26, 2017, at 1:26 PM, merlimat <g...@git.apache.org> wrote: > >> > >> Github user merlimat commented on the issue: > >> > >> https://github.com/apache/incubator-pulsar/issues/523 > >> > >>> If it's configurable then client may start sending msg with > unreasonable size which will be rejected at broker due to > frame-size/storage-limitation, So, client will not know up-front. > >> > >> Max-frame size could be negotiated between client and broker in > Connect/Connected handshake > >> > >> > >> --- > >> If your project is set up for it, you can reply to this email and have > your > >> reply appear on GitHub as well. If your project does not have this > feature > >> enabled and wishes so, or if the feature is enabled but not working, > please > >> contact infrastructure at infrastruct...@apache.org or file a JIRA > ticket > >> with INFRA. > >> --- > > >
