Jaraal wrote: »

TineaCruris wrote: »

They can fix this WITHOUT giving us a cooldown. They proved that during the Midyear Mayem event. That is the problem.

Exactly.

The only real difference between a few years ago and now is that now everybody has shiny, sparkly, glowy particle heavy effects, pets, mounts, and housing items. It's not a coincidence that non-combat pets, mementos, and other such things are disabled in Cyro and IC. They know exactly how much of a load these non-essential things create on the servers. But as long as they are profitable, they will keep selling them in the store.

And rather than improve the infrastructure to handle all these new items, they would rather slow us down (Rapids and Major Expedition nerfs), slow down our Actions Per Minute (increased cooldowns), and otherwise dumb down our game play experience to fit the current hardware and software limitations.

Agree, note that all the sparkle and animated proc sets does not cost the server anything that is outside the proc effect.
It affect the client, and face it most go with the default setting or optimize ESO to look good overland out of combat.
Server lag affect all the same at least in the same fight. Lover the graphic setting in Cyrodil and trials.
You have people having serious issues like game crash porting into populated spots like towns during events or Brenda during midwinter event. And yes we got more sparkle and a lot more styles who has to be loaded into graphic memory.

Yes its server issues, but things stacks. Server lag+ internet lag+lag on your system.
And yes I would have them put the 6 main Cyrodil servers on an 100 core server, pull needed data from database and into memory on it and you could even raise the population cap.

On the other hand before launching Morrowind they lowered the population cap in Reaper march because lots had issues in Rawl. Yes loading all the characters created problems, we got lots more motifs today.

Damnationie wrote: »

Can we just put this "we have to do this on live" nonsense to bed.
What they are currently doing falls into the less then professional end of the IT world.

As a 25 Year Enterprise level IT vet you hit it on the head. Profiling their code on a Production system is doable with bots During a maintenance window. Of course Test systems do not have the same end to end HW. ("hardware" for you noobs) /poke 😁

One would assume @Zos already does this: 🙄
ACE / Content Switches - Attaching Network sniffers and performing data captures.
Routers - Attaching Network sniffers and performing data captures. check marked data types.

Enterprise Firewalls - like Checkpoint firewalls - Review the logs, download to a database and Pivot Table certain "marked" data to see what the offending transactions are.

Servers CPU / Cores- max bandwidth??? PCI lanes

Memory size and speed of course, and if there is any memory overlaid to external storage (bad joojoo)

Servers I/O Channel - You have to have a Server tech that knows WTH they are doing to get this right including doing something most NEVER do. Look at the damn spec sheet of your server and see what your Southbridge I/O bandwidth rating.is. Cooling will affect this as well and will drive it into the dirt if not cooled sufficiently with heavy traffic.

Fiber attached Storage - What good is optically attached storage if your I/O channel on the server is sub par, heats up and performance degrades.

Database - Stats are typically run so doubt they would screw that up. That said there's always some SQL [self removed profanity] setting up a scrape during primetime killing performance, or worse yet someone deciding this is a perfect time to run a Full Server O/S backup across the Network. they'll even say, "It's on a separate VLAN" 🤬😡🤬😡🤬 not understanding how network HW works, thinking everything is Unlimited, /insert Palpitine joke here 😉

Garbage collection...hmmm...

Timefinder on Database or Memory scraping realtime "Session States" so that data consistency is preserved at all times. Can slow things down...depends.

VPN Encryption / Decryption - Shouldn't be a huge issue but needs to be tested every now and then as Vendor Patches sometimes screw things up. Vendor patches can screw up any part of architecture so performance tests need to occur during maintenance windows...............with Bots. 😁

P.S. - If their Datacenter is not their own Datacenter, I.E. a collocation, many of the items I listed above will NOT be done unless hell freezes over. Also the "Towered" nature of IT teams makes it damn near impossible for teams to work together on a problem, as most times they look at their system and say I see no errors, collect paycheck, go home, pay monthly mortgage, say hi to the their wives and my kids. j/k lol 😙😁