Re: Understanding replication
Thank you Alex and Jeff for your response. On Sunday, September 20, 2020, Jeff Jirsa wrote: > Data is always sent to all replicas but is only guaranteed by consistency > level, and ordering into guaranteed if you’re using appropriate serial > consistency levels > > e.g. if you insert into dc1 with LOCAL consistency levels and read from > dc2, dc2’s state is strictly undefined and there is no order implied or > guaranteed > > On Sep 20, 2020, at 9:57 AM, Alex Ott wrote: > > > data is always written to all replicas in the cluster. > here is good diagram how this happens in the multi-dc cluster: > https://docs.datastax.com/en/dse/6.7/dse-arch/datastax_ > enterprise/dbInternals/dbIntClientRequestsMultiDCWrites.html > > regarding the second question - theoretically, yes, it's possible, but if > they were done for the same primary key, then they will be resolved via > data timestamp. > > On Sun, Sep 20, 2020 at 5:31 PM Jai Bheemsen Rao Dhanwada < > jaibheem...@gmail.com> wrote: > >> Hello, >> >> I have a question regarding multi Datacenter replication. >> In a multi datacenter(dc-1, dc-2) if two records are written into dc-1 is >> there a guarantee that these two records replicate to dc-2 in the same >> order or is there a possibility that second insert replicate faster than >> first insert? (Assuming first record is bigger than seconds record in terms >> of packet size). >> > > > -- > With best wishes,Alex Ott > http://alexott.net/ > Twitter: alexott_en (English), alexott (Russian) > >
Re: Understanding replication
Data is always sent to all replicas but is only guaranteed by consistency level, and ordering into guaranteed if you’re using appropriate serial consistency levels e.g. if you insert into dc1 with LOCAL consistency levels and read from dc2, dc2’s state is strictly undefined and there is no order implied or guaranteed > On Sep 20, 2020, at 9:57 AM, Alex Ott wrote: > > > data is always written to all replicas in the cluster. > here is good diagram how this happens in the multi-dc cluster: > https://docs.datastax.com/en/dse/6.7/dse-arch/datastax_enterprise/dbInternals/dbIntClientRequestsMultiDCWrites.html > > regarding the second question - theoretically, yes, it's possible, but if > they were done for the same primary key, then they will be resolved via data > timestamp. > >> On Sun, Sep 20, 2020 at 5:31 PM Jai Bheemsen Rao Dhanwada >> wrote: >> Hello, >> >> I have a question regarding multi Datacenter replication. >> In a multi datacenter(dc-1, dc-2) if two records are written into dc-1 is >> there a guarantee that these two records replicate to dc-2 in the same order >> or is there a possibility that second insert replicate faster than first >> insert? (Assuming first record is bigger than seconds record in terms of >> packet size). > > > -- > With best wishes,Alex Ott > http://alexott.net/ > Twitter: alexott_en (English), alexott (Russian)
Re: Understanding replication
data is always written to all replicas in the cluster. here is good diagram how this happens in the multi-dc cluster: https://docs.datastax.com/en/dse/6.7/dse-arch/datastax_enterprise/dbInternals/dbIntClientRequestsMultiDCWrites.html regarding the second question - theoretically, yes, it's possible, but if they were done for the same primary key, then they will be resolved via data timestamp. On Sun, Sep 20, 2020 at 5:31 PM Jai Bheemsen Rao Dhanwada < jaibheem...@gmail.com> wrote: > Hello, > > I have a question regarding multi Datacenter replication. > In a multi datacenter(dc-1, dc-2) if two records are written into dc-1 is > there a guarantee that these two records replicate to dc-2 in the same > order or is there a possibility that second insert replicate faster than > first insert? (Assuming first record is bigger than seconds record in terms > of packet size). > -- With best wishes,Alex Ott http://alexott.net/ Twitter: alexott_en (English), alexott (Russian)
Understanding replication
Hello, I have a question regarding multi Datacenter replication. In a multi datacenter(dc-1, dc-2) if two records are written into dc-1 is there a guarantee that these two records replicate to dc-2 in the same order or is there a possibility that second insert replicate faster than first insert? (Assuming first record is bigger than seconds record in terms of packet size).
