Which database engine are you using for Zabbix Server?

[xsbr]

Actually we are using MySQL on our Testing Environment (2.5k vps) with this specs:

OS: CentOS 5.5 64bits

Database: MySQL Percona 5.1

• InnoDB with a file per table
• Housekeeper disabled
• Table partition by day on history tables
• Table partition by month on trend tables
• Created an index by clock column on some tables

Hardware:

• Dell R610 Poweredge
• 2x L5520 CPUs
• 24GB RAM
• 4x HDs 500GB SATA 10K (RAID 10)

Now we trying to setup this environment on Oracle Standard 11g

[nated]

DB: MySQL 5.0 (percona), all InnoDB, Master-Master replication running active/passive with manual failover.

Hardware: 2 x Dell 2970, 12 core x 2.8GHz, 32GB, 6 x 15k SAS in RAID10

OS: Debian 5

We're in the process of migrating from the 2970's to a couple R710's with 128GB RAM Also switching to Percona Server 5.1.

I'm interested in reading about some big installs on PostgreSQL. Anyone doing that?

[untergeek]

Now we trying to setup this environment on Oracle Standard 11g

We are presently running 531 hosts, 25409 items, 6491 triggers and 408.98 nvps on RHEL5 with Oracle 11G on the backend. I would be very, very interested in your experience with this. We are slated to grow dramatically with our monitoring soon. The one thing which would dramatically speed up Oracle's performance would be the use of BIND variables. I understand why it's not included in Zabbix's code, but it is a disappointment to our DBAs as every query has to be fully evaluated instead of being able to cache certain elements.

How large is your entire database?

[italux]

How many hosts you have?

DB: MySQL 5.0 (percona), all InnoDB, Master-Master replication running active/passive with manual failover.

Hardware: 2 x Dell 2970, 12 core x 2.8GHz, 32GB, 6 x 15k SAS in RAID10

OS: Debian 5

We're in the process of migrating from the 2970's to a couple R710's with 128GB RAM Also switching to Percona Server 5.1.

I'm interested in reading about some big installs on PostgreSQL. Anyone doing that?

[xsbr]

2.7k hosts - 181k items - 92k triggers - 2471 vps
DB size is about 73GB (2 months trends and 2 days for history)

How large is your entire database?

[jonh]

DB: MySQL 5.1 (stock version), all InnoDB

Hardware: Dell 2970, 8 core x 2.6GHz, 32GB, 2 x 15k SAS in RAID1 (OS, innodb logs) and 4 x 15k SAS in RAID10 (innodb tablespace)

OS: CentOS 5 64-bit

Hosts/Items: 3277 / 133139

Has anyone tested Zabbix on MySQL 5.5? It's finally in public release, but it's new enough that I don't want to just cut over without lots of testing.

Thanks for the comments about partitioning and added indexes. I have not been using those features yet but we've grown to the point where some queries on history and trends literally never complete. So it may be time for that Would you mind saying what clock indexes you added, and why?

[nated]

How many hosts you have?

327 hosts, 22k items, 6k triggers, 336 nvps, and db size around 150GB (lots of trend data). Not that big compared to some of the others here but we are looking to grow it quite a bit. We have some other internal tools using the db also, though they are lightweight compared to Zabbix.

[MarkusL]

Our goal is monitoring small customer networks. We are working with one zabbix-proxy per customer.

At the moment we are migrating with a complete new installation our old 1.6.5-zabbix-server/proxies to a fresh and basic redesigned 1.8.4-zabbix-server/proxy-environtment.

All our proxies connect via outgoing ssh automatically to our zabbix-server.

Some data about our environment:

1x Zabbix-Server:
DELL R2950, 1x XEON Quad E5420 (2,5GHz, 2x6MB), 8GB RAM, 4x 146GB 15k SAS (2x RAID1 for root and swap), 2x 450GB 15k SAS (RAID1 for mysql-innodb),
Ubuntu Server 10.04.1 LTS, 64bit

at the moment 21x Zabbix-Proxy-Server:
BTO-servers, Intel Atom, 1-2GB RAM, 80-160GB S-ATAII,
Ubuntu Server 10.04.1 LTS, 64bit, rollout / cloning via Mondo

Zabbix-Values (when we finished migrating):
~ 600 Hosts, fast growing :-)
~ 16000 Items
~ 20000 Triggers
~ 200vps

Most interesting about the setup is, that the basis of our monitoring is simple "german" T-DSL Business 2000 up to 16000 (upload 40-125kB/s max) with forced reconnect all 24h by provider. We have a "planned downtime" in our trigger-dependency-model for 23.55h-00.05h; in this time we and all our customers Internet go down and up again (router- / firewall-config) + automatically connecting all proxies per autossh to our zabbix-server again. Works with our old system and our new system like a charm for about 2,5 years now!

We really love the model server <-> proxy in zabbix. Btw, all proxies are configured to store config for 48h in case of longer internet-downtimes at us or customer. In this case, proxies store all collected data in local db and deliver at reconnecting. With little tricks in trigger-dep. we only get one alert; no nodata or other things popping up. GREAT!


Kind regards,

Markus.


P.S.: very interesting thread! I´m really interested to hear from other people working with zabbix-proxies.

[xsbr]

Hosts/Items: 3277 / 133139

Great numbers! We're in the same game.

Has anyone tested Zabbix on MySQL 5.5? It's finally in public release, but it's new enough that I don't want to just cut over without lots of testing.

I prefer waiting for percona 5.5 stable version

So it may be time for that Would you mind saying what clock indexes you added, and why?

On next days I'll write about this with details. Today I've only my own documentation in SQL format.

[xsbr]

P.S.: very interesting thread! I´m really interested to hear from other people working with zabbix-proxies.

I opened a dedicated thread to talk about proxies:

http://www.zabbix.com/forum/showthread.php?p=82223

[erikgreen]

Our installation

We're running Zabbix 1.8.4 with:

1212 hosts
40909 Items
16655 Triggers
318.1 items/sec (turned down from 950 by adjusting intervals, more below).


Our hardware:

Web GUI, server: Dell 2950 Dual quad-core Xeon 3.0 Ghz Red Hat Linux 5, 4 GB ram

Database: Oracle, Dual Quad core Xeon 2.3 Ghz, 1U Dell, 8 GB ram, Red Hat Linux, 300GB local disk RAID 1+0 (2 disk stripe mirrored) using Linux software RAID

Oracle has connection pooling enabled for the server, using Oracle 11g.

When I raise the collection rate for items that need a short interval for useful data (like memory and CPU load) beyond about 1 minute, items/sec rises until it nears 600 or so, at which point the server starts to fall behind on its queue. If anything causes a load spike at that point like a report or large graph, performance goes down to an unusable state.

For the most part the database seems ok, we have performance on it tracked tightly with Oracle's tools. It does get IO bound whenever we run a larger report.

I'd be very interested in hearing tuning, setup, and any other info from other oracle users, especially using connection pooling (DRCP).

Erik

[zerg]

1024 hosts(nice number)
44590 items
6238 triggers
200 nvps

FreeBSD 7.2
MySQL 4.1.25
PentiumD E2160
1GB
50Gb NFS on NetApp

[sj7trunks]

This forum is exactly what I was looking for!

Zabbix 1.8.4:
2124 hosts
25237 items ( 7d history / 1y trends )
10746 triggers
277.04 vps

Backend
Oracle 11G
VMWare ESXi

Hardware:
4x @ 3.33GHz
16G Ram
330GB used of 500GB

Problems that I am facing:
1) Disk IO (Who isn't)
2) Disk growth (Oracle is not freeing up space from old rows)
- Investigating table partitioning
3) Performance can be improved by using "BIND" variables (same as untergeek)

My wish list is to get to 100K items like some of the other folk here Good job folks, sharing stats helps others test their own configurations to grow further.

Cheers,
Benjamin

[erikgreen]

I/O config and numbers

Folks -

Can anyone post here their database I/O setup for Zabbix?

We're using Oracle 11.2, direct I/O under Red Hat Linux.

Our disk config is (according to the sysadmins) RAID 1+0, I suspect we're using Linux software RAID on top of SATA disks. We're told we have a two disk stripe for data.

I'm worried about our I/O performance... I'm seeing numbers like load average of 25+ and iowait of 35+ whenever we run a database backup or a large query from Zabbix. Sometimes the data doesn't come up for a graph before 30 seconds expires and it refreshes. We're going to "tune" Oracle to try to help, but I doubt we'll fix this.

Since I'm not a sysadmin here I need some hard information on what storage setup other people are using for Oracle and Zabbix, so I can point our admins in the right direction.

Can you guys please post info? Even mysql configs might be helpful.

Thanks,
Erik

[dotneft]

Number of hosts (monitored/not monitored/templates) 9454 9445 / 0 / 9
Number of items (monitored/disabled/not supported) 1138147 1130648 / 7385 / 114
Number of triggers (enabled/disabled)[problem/unknown/ok] 35828 35828 / 0 [185 / 8724 / 26919]
Number of users (online) 112 3
Required server performance, new values per second 1619.63 -

DB: PostgreSQL 9.0.4
OS: Gentoo with kernel 2.6.39
HW: 2 x Xeon E5420 (8 cores), 10GB FB-DIMM, 2 x SAS HDD (indexes and data), 1 x SATA (system)

PS: PostgreSQL with partitioning.