You are viewing a plain text version of this content. The canonical link for it is here.
Posted to user@ignite.apache.org by summasumma <by...@gmail.com> on 2018/11/29 11:29:49 UTC
Geo redundancy support in Ignite?
Hi all,
Can anyone please let me know if there is Geo redundancy scenario support in
Ignite?
Say, if we need a geo-replication of a ignite cluster, what is the best way
to do/configure? Any Ignite specific suggestions already available?
Can u pls clarify following?
1. Is it possible in ignite to bring up 2 clusters in 2 different
location(geo) and make one cluster as backup for another? meaning, if the
cluster goes down due to some catastrophe in one location, can another
cluster across geo can act as active cluster now? (Active-Standby)
2. Lets say we make 2 server nodes across geo-location as a single cluster,
then the backup replication between them is split. Now when a record is
inserted by client node to this cluster, then half of them will be stored in
the nearby geo location node in cluster and remaining half wil go to the
other server node in the same cluster. So if the client-node does a
key-based lookup 50% of the time it ll come from far geo location and
traverse the network which will create latency?
3. Is there a availablity zone kindof feature in Ignite where we can specify
the client node to communicate with the nearest availaility zone Ignite
cluster after initial negotiation?
Thanks,
...summa
--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/
Re: Geo redundancy support in Ignite?
Posted by Adam Rosini <ar...@jobcase.com>.
We use
https://github.com/apache/ignite/blob/master/modules/core/src/main/java/org/apache/ignite/cache/affinity/rendezvous/ClusterNodeAttributeAffinityBackupFilter.java
in order to say that the backup nodes cannot be in the same availability
zone as the primary node for any given cache. You just need to make sure
each cache has at least 1 backup and your backup count does not exceed your
node count.
On Thu, Nov 29, 2018 at 11:55 AM Stephen Darlington <
stephen.darlington@gridgain.com> wrote:
> It’s absolutely possible. The difficulty comes not in making it work but
> in figuring out the failure scenarios and what sensible thing to do in each
> case. The obvious example: if your inter-site network goes down what
> happens? You now have two clusters, each that thinks it’s *the* cluster.
>
> I don’t mean to discourage you from trying, but, equally, don’t
> underestimate the task.
>
> Regards,
> Stephen
>
> On 29 Nov 2018, at 16:35, Juan Carlos Franzoy <jf...@gmail.com> wrote:
>
> Hello. I am a newbie on Ignite. Ingite does not support geo redundancy out
> of the box. But, looking at the documentation it is not impossible to
> realize it.
>
> Using a mix of multicast and static IO addresses, and good inter-site
> networking, you could establish a geographically distributed Ignite Cluster.
>
> If you redefines AffinityFunction to locate different copies of the same
> key in different sites, you could get that kind of redundancy.
>
> Am I right?
>
> On Thu, 29 Nov 2018 at 09:30, Ilya Kasnacheev <il...@gmail.com>
> wrote:
>
>> Hello!
>>
>> Apache Ignite currently assumes that your whole cluster is geographically
>> collocated, including clients (but excluding thin clients).
>>
>> There may be third party solutions on top of Apache Ignite which allow
>> geo-distributed redundancy.
>>
>> Regards,
>> --
>> Ilya Kasnacheev
>>
>>
>> чт, 29 нояб. 2018 г. в 14:38, summasumma <by...@gmail.com>:
>>
>>> Hi all,
>>>
>>> Can anyone please let me know if there is Geo redundancy scenario
>>> support in
>>> Ignite?
>>>
>>> Say, if we need a geo-replication of a ignite cluster, what is the best
>>> way
>>> to do/configure? Any Ignite specific suggestions already available?
>>>
>>> Can u pls clarify following?
>>> 1. Is it possible in ignite to bring up 2 clusters in 2 different
>>> location(geo) and make one cluster as backup for another? meaning, if the
>>> cluster goes down due to some catastrophe in one location, can another
>>> cluster across geo can act as active cluster now? (Active-Standby)
>>>
>>> 2. Lets say we make 2 server nodes across geo-location as a single
>>> cluster,
>>> then the backup replication between them is split. Now when a record is
>>> inserted by client node to this cluster, then half of them will be
>>> stored in
>>> the nearby geo location node in cluster and remaining half wil go to the
>>> other server node in the same cluster. So if the client-node does a
>>> key-based lookup 50% of the time it ll come from far geo location and
>>> traverse the network which will create latency?
>>>
>>> 3. Is there a availablity zone kindof feature in Ignite where we can
>>> specify
>>> the client node to communicate with the nearest availaility zone Ignite
>>> cluster after initial negotiation?
>>>
>>> Thanks,
>>> ...summa
>>>
>>>
>>>
>>> --
>>> Sent from: http://apache-ignite-users.70518.x6.nabble.com/
>>>
>>
>
> --
> Juan Carlos Franzoy
> Licenciado en Sistemas
>
>
>
>
--
Adam Rosini, Software Engineer
Jobcase, Inc. <https://www.jobcase.com/>| 201 Broadway | 7th Floor |
Cambridge, MA 02139
arosini@jobcase.com
Disclaimer
The information contained in this communication from the sender is confidential. It is intended solely for use by the recipient and others authorized to receive it. If you are not the recipient, you are hereby notified that any disclosure, copying, distribution or taking action in relation of the contents of this information is strictly prohibited and may be unlawful.
This email has been scanned for viruses and malware, and may have been automatically archived by Mimecast Ltd, an innovator in Software as a Service (SaaS) for business. Providing a safer and more useful place for your human generated data. Specializing in; Security, archiving and compliance. To find out more visit the Mimecast website.
Re: Geo redundancy support in Ignite?
Posted by Stephen Darlington <st...@gridgain.com>.
It’s absolutely possible. The difficulty comes not in making it work but in figuring out the failure scenarios and what sensible thing to do in each case. The obvious example: if your inter-site network goes down what happens? You now have two clusters, each that thinks it’s *the* cluster.
I don’t mean to discourage you from trying, but, equally, don’t underestimate the task.
Regards,
Stephen
> On 29 Nov 2018, at 16:35, Juan Carlos Franzoy <jf...@gmail.com> wrote:
>
> Hello. I am a newbie on Ignite. Ingite does not support geo redundancy out of the box. But, looking at the documentation it is not impossible to realize it.
>
> Using a mix of multicast and static IO addresses, and good inter-site networking, you could establish a geographically distributed Ignite Cluster.
>
> If you redefines AffinityFunction to locate different copies of the same key in different sites, you could get that kind of redundancy.
>
> Am I right?
>
> On Thu, 29 Nov 2018 at 09:30, Ilya Kasnacheev <ilya.kasnacheev@gmail.com <ma...@gmail.com>> wrote:
> Hello!
>
> Apache Ignite currently assumes that your whole cluster is geographically collocated, including clients (but excluding thin clients).
>
> There may be third party solutions on top of Apache Ignite which allow geo-distributed redundancy.
>
> Regards,
> --
> Ilya Kasnacheev
>
>
> чт, 29 нояб. 2018 г. в 14:38, summasumma <bytestreams@gmail.com <ma...@gmail.com>>:
> Hi all,
>
> Can anyone please let me know if there is Geo redundancy scenario support in
> Ignite?
>
> Say, if we need a geo-replication of a ignite cluster, what is the best way
> to do/configure? Any Ignite specific suggestions already available?
>
> Can u pls clarify following?
> 1. Is it possible in ignite to bring up 2 clusters in 2 different
> location(geo) and make one cluster as backup for another? meaning, if the
> cluster goes down due to some catastrophe in one location, can another
> cluster across geo can act as active cluster now? (Active-Standby)
>
> 2. Lets say we make 2 server nodes across geo-location as a single cluster,
> then the backup replication between them is split. Now when a record is
> inserted by client node to this cluster, then half of them will be stored in
> the nearby geo location node in cluster and remaining half wil go to the
> other server node in the same cluster. So if the client-node does a
> key-based lookup 50% of the time it ll come from far geo location and
> traverse the network which will create latency?
>
> 3. Is there a availablity zone kindof feature in Ignite where we can specify
> the client node to communicate with the nearest availaility zone Ignite
> cluster after initial negotiation?
>
> Thanks,
> ...summa
>
>
>
> --
> Sent from: http://apache-ignite-users.70518.x6.nabble.com/ <http://apache-ignite-users.70518.x6.nabble.com/>
>
>
> --
> Juan Carlos Franzoy
> Licenciado en Sistemas
Re: Geo redundancy support in Ignite?
Posted by Juan Carlos Franzoy <jf...@gmail.com>.
Hello. I am a newbie on Ignite. Ingite does not support geo redundancy out
of the box. But, looking at the documentation it is not impossible to
realize it.
Using a mix of multicast and static IO addresses, and good inter-site
networking, you could establish a geographically distributed Ignite Cluster.
If you redefines AffinityFunction to locate different copies of the same
key in different sites, you could get that kind of redundancy.
Am I right?
On Thu, 29 Nov 2018 at 09:30, Ilya Kasnacheev <il...@gmail.com>
wrote:
> Hello!
>
> Apache Ignite currently assumes that your whole cluster is geographically
> collocated, including clients (but excluding thin clients).
>
> There may be third party solutions on top of Apache Ignite which allow
> geo-distributed redundancy.
>
> Regards,
> --
> Ilya Kasnacheev
>
>
> чт, 29 нояб. 2018 г. в 14:38, summasumma <by...@gmail.com>:
>
>> Hi all,
>>
>> Can anyone please let me know if there is Geo redundancy scenario support
>> in
>> Ignite?
>>
>> Say, if we need a geo-replication of a ignite cluster, what is the best
>> way
>> to do/configure? Any Ignite specific suggestions already available?
>>
>> Can u pls clarify following?
>> 1. Is it possible in ignite to bring up 2 clusters in 2 different
>> location(geo) and make one cluster as backup for another? meaning, if the
>> cluster goes down due to some catastrophe in one location, can another
>> cluster across geo can act as active cluster now? (Active-Standby)
>>
>> 2. Lets say we make 2 server nodes across geo-location as a single
>> cluster,
>> then the backup replication between them is split. Now when a record is
>> inserted by client node to this cluster, then half of them will be stored
>> in
>> the nearby geo location node in cluster and remaining half wil go to the
>> other server node in the same cluster. So if the client-node does a
>> key-based lookup 50% of the time it ll come from far geo location and
>> traverse the network which will create latency?
>>
>> 3. Is there a availablity zone kindof feature in Ignite where we can
>> specify
>> the client node to communicate with the nearest availaility zone Ignite
>> cluster after initial negotiation?
>>
>> Thanks,
>> ...summa
>>
>>
>>
>> --
>> Sent from: http://apache-ignite-users.70518.x6.nabble.com/
>>
>
--
Juan Carlos Franzoy
Licenciado en Sistemas
Re: Geo redundancy support in Ignite?
Posted by Ilya Kasnacheev <il...@gmail.com>.
Hello!
Apache Ignite currently assumes that your whole cluster is geographically
collocated, including clients (but excluding thin clients).
There may be third party solutions on top of Apache Ignite which allow
geo-distributed redundancy.
Regards,
--
Ilya Kasnacheev
чт, 29 нояб. 2018 г. в 14:38, summasumma <by...@gmail.com>:
> Hi all,
>
> Can anyone please let me know if there is Geo redundancy scenario support
> in
> Ignite?
>
> Say, if we need a geo-replication of a ignite cluster, what is the best way
> to do/configure? Any Ignite specific suggestions already available?
>
> Can u pls clarify following?
> 1. Is it possible in ignite to bring up 2 clusters in 2 different
> location(geo) and make one cluster as backup for another? meaning, if the
> cluster goes down due to some catastrophe in one location, can another
> cluster across geo can act as active cluster now? (Active-Standby)
>
> 2. Lets say we make 2 server nodes across geo-location as a single cluster,
> then the backup replication between them is split. Now when a record is
> inserted by client node to this cluster, then half of them will be stored
> in
> the nearby geo location node in cluster and remaining half wil go to the
> other server node in the same cluster. So if the client-node does a
> key-based lookup 50% of the time it ll come from far geo location and
> traverse the network which will create latency?
>
> 3. Is there a availablity zone kindof feature in Ignite where we can
> specify
> the client node to communicate with the nearest availaility zone Ignite
> cluster after initial negotiation?
>
> Thanks,
> ...summa
>
>
>
> --
> Sent from: http://apache-ignite-users.70518.x6.nabble.com/
>