Time |
S |
Nick |
Message |
06:10 |
|
|
Haudegen joined #rest |
07:07 |
|
|
interop_madness joined #rest |
07:07 |
|
|
interop_madness joined #rest |
07:57 |
|
|
zinox joined #rest |
09:20 |
|
|
interop_madness joined #rest |
09:20 |
|
|
interop_madness joined #rest |
10:15 |
|
|
riddle joined #rest |
10:15 |
|
|
`0660_ joined #rest |
10:17 |
|
|
everbrew_ joined #rest |
11:42 |
|
|
Haudegen joined #rest |
12:05 |
|
|
donofrio joined #rest |
15:28 |
|
|
realnot joined #rest |
16:02 |
|
|
wsieroci joined #rest |
16:41 |
|
|
Guest71929 joined #rest |
20:52 |
|
|
ModusPwnens joined #rest |
20:54 |
|
ModusPwnens |
I've run into an interesting situation where there are multiple different data stores for the same type of resource (e.g. short lived objects that need to be stored in one of two different redis clusters). In REST, how can I ensure that I know which resource IDs correspond with which redis cluster? I really don't like the idea of storing a mapping somewhere. |
20:56 |
|
|
Haudegen joined #rest |
21:56 |
|
whartung |
ModusPwnens: You have to either be able to look it up by the resource ID, or calculate it from the resource ID, there’s not alot of other options. |
21:56 |
|
whartung |
there’s nothing that say the resources you publish match the ones on your backend. |
21:56 |
|
whartung |
so you can always make somethin gup |
21:59 |
|
ModusPwnens |
i've been doing some more digging |
21:59 |
|
ModusPwnens |
i think i can just hash the resource ID and then do a modulo N peration, where N is the number of our redis clusters |
22:00 |
|
ModusPwnens |
alternatively, I can hash a POJO into a resource ID. POJO could be something like {clusterId: <clusterId>, redisKey: <redisKey>} and I could base64 it into a string and use that as the ID |
22:01 |
|
whartung |
sure, taht can always work — as long as the resources don’t move. |
22:05 |
|
ModusPwnens |
yeah, for us the resources don't move |
22:05 |
|
ModusPwnens |
if they did, we'd have to use more of a consistent hashing ring concept |
22:06 |
|
whartung |
then just tack the redis id on the end of the resource id and call it good. |
22:07 |
|
whartung |
123456_1 |
22:07 |
|
whartung |
hashing doesn’t help, as you can’t “unhash” |
22:07 |
|
whartung |
just make a composite ID |