Hi Alex,
What memory hit rate I can see is irrelevant. That is entirely algorithm
dependent. What is important is that object re-access patterns more
closely follow real-world scenarios, rather than some arbitrary forced
memory-miss scenario.
This then allows for better memory caching algorithms to strut their stuff,
while others may only be capable of much less. The point is, if the load
follows more real-world patterns, then caches that are tuned for that
behaviour can perform their intended function, rather than being reduced
to a disk-only test.
Stew.
At 06:04 PM 11/30/99 -0700, Alex Rousskov wrote:
>Stew,
>
> You forgot to include the memory hit ratios that you are seeing
>(or those that your research shows). We do not want to just restrict
>memory hit ratios per se, we want to see realistic numbers that you are
>seeing. Unfortunately, even your fairly detailed message is missing
>those numbers so we are left with no positive info thatwe can use.
>
>Please provide us with the measurements you see in practice, and we will
>try to adjust the workload accordingly (given that other vendor's
>numbers or expectations match yours to a certain extent). If we do not
>have any numbers, we have to take educated guesses and be on the safe
>side. The latter is not the best approach, of course.
>
>Thanks,
>
>Alex.
>
>On Tue, 30 Nov 1999, Stew Forster wrote:
>
>> Why 4-6% memory hit ratio? Where has this figure come from? How does
>> this relate to real world behaviour? Some vendors have spent a lot of time
>> explicity studying real world logs and running simulations and tuning algorithms
>> to handle real-world situations. These correspond directly to improved
>> memory hit ratios.
>>
>> I understand the need to prevent the entire working set fitting into RAM, but
>> lets face it. If vendor X can get 1000TPS say, then at ~13KB/object and a
>> 55% hit rate, we're talking about 5.85MB/sec of new object data flowing into
>> the system, or 170 seconds per GB. There's no way any working set will
>> fit into RAM in those circumstances.
>>
>> Further, our research shows that for 1 week's worth of traffic access at 200TPS,
>> 75% of all cache hits occur within 7 hours of an object being pulled in, and indeed
>> 56% within the first hour. Our research also shows that as TPS increases,
>> the percentages increase, meaning that each object is more likely to be accessed
>> more quickly. This correlates directly to higher achievable RAM hit rates.
>>
>> By artificially restricting everyone to 4-6% memory hits you're stating that vendor X's
>> approach is invalid if vendor X focussed on improved memory performance.
>>
>> Stew.
>
This archive was generated by hypermail 2b29 : Tue Jul 10 2001 - 12:00:09 MDT