R81.10 MGMT to VSX

[Magnus-Holmberg]

Hi,

After upgrading our MDS to R81.10 we see a significant improvment of the speed of adding routes / deploying VS on VSX system.
We have some VSX Clusters with 50 VS and a few hundred routes, before it took ages to change route in these (10min+)
This is now down to 1min! 😄 (VSX are still on R80.30 3.10)
Same result on multiple VSX and MDS clusters 🙂

Have anyone experiance similar improvments?

Regards,
Magnus

[Kaspars_Zibarts]

Not yet.. still on R80.40 on both: MGMT and VSX. But we haven't seen that bad performance adding routes. Ok, it's not 1min but certainly not 10! But we only have 20 VS max on a single cluster 🙂 though with hundreds of routes

[genisis__]

Still on R80.40 as well, but after the last set of fixes (dragging R&D into calls), our platform  extremely stable.

From a service prospective, I see no driver at this point to upgrade.

That said, will be working on a green field installation shortly with new tin, this will be installed with R81.10 running VSX.

[Magnus-Holmberg]

from our experience we have a massive increase of deployment time after 25VS.

1-20 is normally quick 🙂

[Alex-]

From a management standpoint alone, R81/R81.10 is worth the upgrade as the speed improvements are really noticeable. Also, bonuses which are appreciated by the customers like NAT hit count, Web Smart Console, updateable Smart Console and concurrent policy installation.

Now on VSX, I have to say that R81 and R81.10 are really stable. With R80.40, I had recurrent routed issues requiring private hotfixes or tweaking which went away after upgrade.

Accelerated policy installation is also a nice to have on the long run in the times when a policy is fine-tuned, requiring multiple policy installations in a row.

My first R81 upgrade went bad but it turned out to be a series of external dependencies, but top job from Check Point for this new generation of Gaia OS.

[genisis__]

I've had a stab at upgrading our R80.30 MDS to R81.10 but had to back out (TAC involved).  Will be attempting this again in the next few months.

Its nice to hear positive 'real' world feed back for VSX as well, did you use dynamic balancing as well?  If so how have you addressed monitoring individual VS's? ie. CPU utilisation on a per VS basis?

[Alex-]

No, I don't use dynamic split yet. I can't say for MDS, I use only SMS in my various implementations and no issues there. I upgraded from R80.40 with the Blink images.

When I speak of R80.40, I don't want to be dramatic. 😉

I had issues with the routed daemon in several installations to the point it was always a point of attention after deployments but yes the systems were generally running fine.