OSCAM Hashed Tables (beta code-release)

[CapNCooK]

I am not using google translate.
"Fashion" is only expression, I've exchange cacheex-3 with many peers and since yesterday lot of my peers asked me for switching to cacheex-2 and camd35. For me that is like new Fashion.

I seriously believed that google translated "Mode" into "Fashion" :)

Anyways,

For camd35 is good idea because is fast.

True, but no connection control since its UDP. I would generally recommend to use cs378x.

Cacheex 2 is designed to use mixed requests between normal and cache exchange, probably is good for beginners or people with limited internet connection and small amount of peers.

In SVN .. yes., because mode-2 is broken on many places.
In HT .. no., you could just use it to push cache without requests. (cacheex_allow_request = 0)
When combined, it works.. but not recommended because of the ecm pending tables.

Conclusions:
There is no idea to cacheex 2 switching when you have a loot of peers and datacenter based server.

True!

Also, to all, take a close look to blueven's release notes. Really a lot is new/changed.

[galiano]

.....Filtering best csp craps or multics cache hwo are not good.......

In this vein, I strongly hope in a later builds, we see option "cacheex_drop_multics". To avoid having we manually delete mcs peers.

[turbopower]

I seriously believed that google translated "Mode" into "Fashion" :)

Anyways,

True, but no connection control since its UDP. I would generally recommend to use cs378x.


In SVN .. yes., because mode-2 is broken on many places.
In HT .. no., you could just use it to push cache without requests. (cacheex_allow_request = 0)
When combined, it works.. but not recommended because of the ecm pending tables.


True!

Also, to all, take a close look to blueven's release notes. Really a lot is new/changed.

Wow Cap you surprising me every time :) Yes you are right on Bulgarian - Fashion is translated like Moda. Now I understand what you mean about google translate :) But any way thank you about this, I will always remember that you are the only one non Bulgarian person ever discussed in Bulgarian section here.

Because you know in details what exactly is implemented from blueven, could you please give us advise what is the best setup.

[turbopower]

In this vein, I strongly hope in a later builds, we see option "cacheex_drop_multics". To avoid having we manually delete mcs peers.

Please not compare mcs cache with msc cacheex. First one is internal cache algorithm for msc but second one is designed from and for oscam cacheex and is not push fake cw in any case.

[gimic]

@gimic
you look for small ping above 100ms or a good and stable CS?

Some one tell me on PM i'm scaming here-because he want's all cache with stable ping.But this is other point of view.
We don't need huge amounth-using mode 3 we have on cache 10 times or 100 times more the same dw we nee 1-and if is into we need to provide to peer.
Now all cache decrease but increase hits and is more stabble because charging of processor come down.

Example: a lot of people tell me -i want to see ping under 80 ms from 09cd card?
For me are good under 642 ms all dw because under this time 09cd work's fine without freeze.
It's is a stupidity---im my point of view.
latency= time from cache+time from card+time from net if are above a 500 ms value for 09 card and above 90 for 17 it sems we can watch tv without freeze.
I watch a channel using dw above 250 ms and my friend using a 80 ms. This don't speed up time to watch. The moviea beggins and end at the same hour for both of us :)

i agree with your statement of latancy, i don't have such setup running, i was just wondering if it worked that way.

[CapNCooK]

In this vein, I strongly hope in a later builds, we see option "cacheex_drop_multics". To avoid having we manually delete mcs peers.

Its undetectable. CSP/MCS cache looks the same. Plus, MCS cache is usable, so why block it ?

[galiano]

Its undetectable. CSP/MCS cache looks the same. Plus, MCS cache is usable, so why block it ?

I did a simple test:
Two cache account with approximately equal cache size, only to 1702, one line of mcs and one of oscam. Both accounts are being tested separately.

First, mcs line to oscam: waiting 17:800 resulting zero hits. Then increased the time to wait 17:1500 resulting 31% hits
Second, oscam to oscam: waiting 17:800 resulting 83% hits. Then increased the time to wait 17:1500 resulting 98% hits

[turbopower]

I did a simple test:
Two cache account with approximately equal cache size, only to 1702, one line of mcs and one of oscam. Both accounts are being tested separately.

First, mcs line to oscam: waiting 17:800 resulting zero hits. Then increased the time to wait 17:1500 resulting 31% hits
Second, oscam to oscam: waiting 17:800 resulting 83% hits. Then increased the time to wait 17:1500 resulting 98% hits

This is mcs cacheex or mcs internal cache compared to oscam cacheex?

I am not sure how it's working mcs cacheex because there is no enough information about this and as you know mcs is closed source, but for me the important part is as Cap mentioned - the cacheex from mcs is usable. Please don't expect to be productive like oscam one.

Below is described the same point of view from mcs perspective.

I have Total Hits = 27895 from incoming oscam cacheex but

Total Hits = 198676 from internal mcs cache.

From this (mcs) point of view oscam cachex is not effective as internal one if I am thinking in your way.
But this is not true and you are not right.

For me any extra hits are usable if the different camds are not messing up each other.

[CapNCooK]

The new code has been submitted to SVN, therefore i close this thread.
Thanks for all feedback!.