Is it time for another round of thread view count requests?

Forums › Geekzone › Is it time for another round of thread view count requests?

kenkeniff

628 posts

Ultimate Geek

#150942 8-Aug-2014 12:24

Registered users at least would be nice.

"Eventual consistency" even would be ok.

Dumb increments (counts F5) would be better than nothing (seems to work for TradeMe lol).

Please.

1 | 2

6614 posts

Uber Geek
Inactive user

#1104823 8-Aug-2014 12:25

Lol, Good old mash F5 trademe. I setup a few VM's and let them run overnight refreshing 5 times a second and lol. had close to 2 million views by the morning.

kenkeniff

628 posts

Ultimate Geek

#1104827 8-Aug-2014 12:33

TimA: Lol, Good old mash F5 trademe. I setup a few VM's and let them run overnight refreshing 5 times a second and lol. had close to 2 million views by the morning.

Doh, that's not going to fill freitasm with confidence..

freitasm

BDFL - Memuneh
79250 posts

Uber Geek

Administrator

ID Verified

Trusted

Geekzone

Lifetime subscriber

#1104834 8-Aug-2014 12:48

One of the reasons I removed it was for speed - less database updates, especially in very busy threads.

Not seeing it coming back yet. I am tweaking with the idea of changing something in our database to improve performance, might come back then.

gzt

17104 posts

Uber Geek

Lifetime subscriber

#1104837 8-Aug-2014 12:52

I have to admit being morbidly curious about the reason for removing the feature (long ago). Was it commercial sensitivity?

Edit: Posted this prior to above post arriving. I consider it answered.

freitasm

BDFL - Memuneh
79250 posts

Uber Geek

Administrator

ID Verified

Trusted

Geekzone

Lifetime subscriber

#1104838 8-Aug-2014 12:53

Nope, purely database performance (index locks, row locks, etc).

kenkeniff

628 posts

Ultimate Geek

#1104840 8-Aug-2014 12:54

freitasm: One of the reasons I removed it was for speed - less database updates, especially in very busy threads.

Not seeing it coming back yet. I am tweaking with the idea of changing something in our database to improve performance, might come back then.

Is it database I/O or specific row/table locking you're concerned about (hence the latter I suggested eventual consistency so don't need to queue up write locks on a single row/table)?

EDIT: Saw your post above. In that case eventual consistency by writing views to cache then lazily updating a main counter will solve this (the Youtube's etc do this).

freitasm

BDFL - Memuneh
79250 posts

Uber Geek

Administrator

ID Verified

Trusted

Geekzone

Lifetime subscriber

#1104846 8-Aug-2014 13:10

Not I/O, but locking strategy. We currently have very low number of I/O interruptions because almost the entire database is in memory. Transactrion Log is the one thing that consumes I/O mostly, so not a big deal.

We did change to READ COMMITTED SNAPSHOT a while back to improve locking speed and prevent waits.

Equinox IT

Cloud spending continues to surge globally, but most organisations haven’t made the changes necessary to maximise the value and cost-efficiency benefits of their cloud investments. Download the whitepaper From Overspend to Advantage now.

gzt

17104 posts

Uber Geek

Lifetime subscriber

#1104859 8-Aug-2014 13:31

freitasm: One of the reasons I removed it was for speed - less database updates, especially in very busy threads.

Would maintaining count only for registered user views be any more feasible in that regard?

freitasm

BDFL - Memuneh
79250 posts

Uber Geek

Administrator

ID Verified

Trusted

Geekzone

Lifetime subscriber

#1104867 8-Aug-2014 13:43

Would be terrible undercounted. The number of visits from non-registered or non-logged in users is much bigger.

kenkeniff

628 posts

Ultimate Geek

#1104870 8-Aug-2014 13:45

freitasm: Would be terrible undercounted. The number of visits from non-registered or non-logged in users is much bigger.

Still useful though.

Thread XYZ (999 Registered Views, 0 Replies)

gzt

17104 posts

Uber Geek

Lifetime subscriber

#1104874 8-Aug-2014 13:52

I agree, but it would give good indication of community/interest engagement. If counting the other group becomes possible that can be added also. The two figures are complementary and both give useful information.

MikeB4

18435 posts

Uber Geek

ID Verified

Trusted

#1104891 8-Aug-2014 14:23

The last (big) forum I was Admin for we did away with page views as the hit on performance was such it did not justify keeping for very little benefit, we did away with it.

kenkeniff

628 posts

Ultimate Geek

#1104892 8-Aug-2014 14:27

KiwiNZ: The last (big) forum I was Admin for we did away with page views as the hit on performance was such it did not justify keeping for very little benefit, we did away with it.

Cool story, neither assumption applies here

MikeB4

18435 posts

Uber Geek

ID Verified

Trusted

#1104912 8-Aug-2014 14:48

kenkeniff:
KiwiNZ: The last (big) forum I was Admin for we did away with page views as the hit on performance was such it did not justify keeping for very little benefit, we did away with it.

Cool story, neither assumption applies here

It doesn't ?

gzt

17104 posts

Uber Geek

Lifetime subscriber

#1104923 8-Aug-2014 15:05

KiwiNZ:
kenkeniff:
KiwiNZ: The last (big) forum I was Admin for we did away with page views as the hit on performance was such it did not justify keeping for very little benefit, we did away with it.

Cool story, neither assumption applies here

It doesn't ?

It was exactly the reason freitasm removed it, so it seems a similar performance experience applies.

Please guys don't start a he said/i said and derail the thread. Just leave it there please.

1 | 2