How we created SPAE - SaaS for administrators of Internet servers

How did it all start?

Once, two administrators, with sufficient experience in administering zoo servers, thought about the fact that for each server environment you have to configure, or historically they already have, their own nagios, mrtg, cacti, zabbix, and nessus and etc ... schedules, service monitoring and security services is a trivial task, but still this task takes time, requires the reproduction of entities and the construction of new dependencies. It was a logical idea to make “for yourself” a tool that would help in the routine administration as much as possible. But before “inventing a new bicycle” you need to decompose the problem that it will solve into its components.

What is the actual problem?

1. I / the client has only one server. But in order to monitor it you need another one.
And who will monitor the server that monitors this server? :)
There is an exit! Google will help you find a pack of bourgeois services. Tried 15 pieces and did not find one
which would work as we would like and had sufficient functionality.
')
2. In order to put the server on monitoring, it is necessary to register it at least in several configs.
In fact, it is not difficult to run vi config.cfg n times and download the software, but if you need to install or update the software, this creates an additional headache.

H. All tools in different interfaces.
I will give an example of the browser average administrator:
The first tab is nagios, the second is cacti / mrtg. Plus, to check the security is still some kind of software.
These are three different interfaces, but they are connected by one entity.

4. I want to receive notifications through real SMS.
Operators mail gateways either often have a large delay or they simply do not exist.
We must love to build relationships with operators. Either screw the phone to the server,
or work through which thread to the sms-cantor, having figured out its self-written API.

5. Forget s to conduct server security audits.
There are very few such web services and they do not always work efficiently.
But there is a convenient and free nessus. Although not automated without a normal web interface.

What Admin Want?

Based on their own experiences and desires, a list of functionality and features has appeared:

• Simplify adding a server to the monitoring system. It is enough to install from the ports of net-snmpd and with the mouse in putty to build the generated config.
• Inventory - A single place for storing information about what kind of system it is and what hardware it lives on.
• Monitoring the availability of services and metrics: HTTP, PING, SMTP, SSH, POP3, IMAP, DNS, CPU Usage, Disk Spae.
• Notifications by email, real (instant) sms, icq. Daily summary report.
• Resource Usage Charts - Interface, CPU Usage, LoadAvarage, Memory, Disks
• Security - external server check for all known vulnerabilities - frontend for nessus

Development.

There were no free hands for programmers, so I had to learn how to program and write myself.
Java was chosen as the web-frontend as the ajax framework of ZKoss.
AppServer - Tomcat. Web bundle - Apache + Nginx.
Backend - nagios, for rrdtool graphs, for security - nessus, everything is connected by ~ 20 demons and scripts in C, Shell / AWK, Python. After 6 months, the world saw the alpha version.

What happened.

Monitoring:



Download schedules:


Security Interface:


Inventory:



You can kill the demo:

user: demo@shalb.com
password: shalb
spae.shalb.com

Promo page: shalb.com/en/spae/spae_features

Business.

Since it was just silly to spend so much time without the option to earn, registration was added
and billing. We made an analysis of prices for at least some similar projects (mon.itor.us, host-tracker), calculated the cost price of the service for monitoring one server, built a pricing policy and released $ 5 a month for monitoring one IP.

Who is a customer

SPAE's target clients are the owners, administrators of servers, VPS, which have from one to five servers.
We do not focus on Guru administrators and administrators of large server systems. These people have everything for a long time.
The service simply solves some of the problems associated with the administration and security of servers.

The present.

Now the system monitors 80 servers in 5 data centers.
An affiliate program was launched with Volya data center, which gives a 20% discount on our service.
We started refactoring the code with real Java programmers. We catch bugs, we plan a marketing campaign.

Future:

Custom setting of service monitoring (by ports / advanced http monitoring)
Custom OID for graphs (disk IO inclusive)
Multilingual interface.
My IPS - integration into the snort management and notification interface.
WebServices API.
OpenSource standalone version - as soon as the code is not ashamed to show the community.

Thank.