Do we need more RAM?

[TheBigK]

And just to note...not sure if you have a dedicated or VPS or what you're using with Liquidweb (@TheBigK ) but you have about 1000 people online right now (mostly guests) and your site is blazing fast for me. I'm not exactly sure that 3AM my time (Pacific) is exactly a peak period for your site, but it certainly feels really quick, for having that many people on the site.

It's a VPS and the site's fast because we've LiteSpeed server. We now load a ton of JS; but glad to know that the site's loading fast. It's not peak for us; but yeah that's the average traffic.

 

[TPerry]

I have to agree with @WSWD, for 1100 users (almost) online currently your site does not appear to be struggling in the least. A lot of people would kill for that speed.

 

[Floren]

Nah that's impossible.
ES alone takes 512MB to 2GB of ram with your post.
And for 1.7M post you should allocate more than 2GB of ram for Innodb Engine.
So I don't think what you said is true, sorry.

Is not impossible, is how you optimize your server.



Elasticsearch will use whatever it needs for RAM, as well the other processes. From the screenshot, I allocated 1G to Elastic indexes, so that means the rest of server processes run on 60M?
I also have 3 ramdisks, barely used:

#  df -ahT
Filesystem           Type         Size  Used Avail Use% Mounted on
/dev/mapper/vg_os-lv_root
                     ext4         9.9G  1.7G  7.7G  18% /
proc                 proc            0     0     0    - /proc
sysfs                sysfs           0     0     0    - /sys
devpts               devpts          0     0     0    - /dev/pts
tmpfs                tmpfs        3.8G     0  3.8G   0% /dev/shm
/dev/md0             ext4         496M  123M  348M  27% /boot
/dev/mapper/vg_os-lv_home
                     ext4         9.9G  202M  9.2G   3% /home
/dev/mapper/vg_os-lv_opt
                     ext4         9.9G  151M  9.2G   2% /opt
/dev/mapper/vg_os-lv_tmp
                     ext4         5.0G  139M  4.6G   3% /tmp
/dev/mapper/vg_os-lv_var
                     ext4          79G  2.2G   73G   3% /var
tmpfs                tmpfs        8.0M     0  8.0M   0% /var/lib/nginx/client
tmpfs                tmpfs        512M   16K  512M   1% /var/lib/nginx/fastcgi
tmpfs                tmpfs        1.0G     0  1.0G   0% /var/run/mysqld/tmp
none                 binfmt_misc     0     0     0    - /proc/sys/fs/binfmt_misc
/etc/named           none         9.9G  1.7G  7.7G  18% /var/named/chroot/etc/named
/var/named           none          79G  2.2G   73G   3% /var/named/chroot/var/named
/etc/named.conf      none         9.9G  1.7G  7.7G  18% /var/named/chroot/etc/named.conf
/etc/named.rfc1912.zones
                     none         9.9G  1.7G  7.7G  18% /var/named/chroot/etc/named.rfc1912.zones
/etc/rndc.key        none         9.9G  1.7G  7.7G  18% /var/named/chroot/etc/rndc.key
/usr/lib64/bind      none         9.9G  1.7G  7.7G  18% /var/named/chroot/usr/lib64/bind
/etc/named.iscdlv.key
                     none         9.9G  1.7G  7.7G  18% /var/named/chroot/etc/named.iscdlv.key
/etc/named.root.key  none         9.9G  1.7G  7.7G  18% /var/named/chroot/etc/named.root.key

 

[rdn]

Is not impossible, is how you optimize your server.

View attachment 76564

Elasticsearch will use whatever it needs for RAM, as well the other processes. From the screenshot, I allocated 1G to Elastic indexes, so that means the rest of server processes run on 60M?
I also have 3 ramdisks, barely used:

#  df -ahT
Filesystem           Type         Size  Used Avail Use% Mounted on
/dev/mapper/vg_os-lv_root
                     ext4         9.9G  1.7G  7.7G  18% /
proc                 proc            0     0     0    - /proc
sysfs                sysfs           0     0     0    - /sys
devpts               devpts          0     0     0    - /dev/pts
tmpfs                tmpfs        3.8G     0  3.8G   0% /dev/shm
/dev/md0             ext4         496M  123M  348M  27% /boot
/dev/mapper/vg_os-lv_home
                     ext4         9.9G  202M  9.2G   3% /home
/dev/mapper/vg_os-lv_opt
                     ext4         9.9G  151M  9.2G   2% /opt
/dev/mapper/vg_os-lv_tmp
                     ext4         5.0G  139M  4.6G   3% /tmp
/dev/mapper/vg_os-lv_var
                     ext4          79G  2.2G   73G   3% /var
tmpfs                tmpfs        8.0M     0  8.0M   0% /var/lib/nginx/client
tmpfs                tmpfs        512M   16K  512M   1% /var/lib/nginx/fastcgi
tmpfs                tmpfs        1.0G     0  1.0G   0% /var/run/mysqld/tmp
none                 binfmt_misc     0     0     0    - /proc/sys/fs/binfmt_misc
/etc/named           none         9.9G  1.7G  7.7G  18% /var/named/chroot/etc/named
/var/named           none          79G  2.2G   73G   3% /var/named/chroot/var/named
/etc/named.conf      none         9.9G  1.7G  7.7G  18% /var/named/chroot/etc/named.conf
/etc/named.rfc1912.zones
                     none         9.9G  1.7G  7.7G  18% /var/named/chroot/etc/named.rfc1912.zones
/etc/rndc.key        none         9.9G  1.7G  7.7G  18% /var/named/chroot/etc/rndc.key
/usr/lib64/bind      none         9.9G  1.7G  7.7G  18% /var/named/chroot/usr/lib64/bind
/etc/named.iscdlv.key
                     none         9.9G  1.7G  7.7G  18% /var/named/chroot/etc/named.iscdlv.key
/etc/named.root.key  none         9.9G  1.7G  7.7G  18% /var/named/chroot/etc/named.root.key

For a smaller site, yes.

 

[Floren]

For a smaller site, yes.

You mean the memory usage will increase if you have 500 guests online? It will, but it will not double or triple.
Is very easy to test it, run Siege. Here's some proof, I just ran a test with 1000 concurrent users on my little server. Server load and memory during the 1000 concurrent users:



Siege results:

 

[rdn]

I know we both cache pages for the guest

 

[Floren]

I know we both cache pages for the guest

I disabled the cache for this test, or else I would not get 6 seconds to load a page. But you are saying the memory will increase exponentially with the number of users, which is not true. The server load will go to the sky ya, not the memory usage.