I've designed and helped implementing the following
production systems, all of which are still in service:
SugarLabs's Internet infrastructure — I'm coordinator
of the Infrastructure Team,
taking care of 8 machines hosted at various colocation facilities, running
development and support services for a large community.
Free Software Foundation's virtualization infrastructure —
I've been the fall system administrator intern in 2009. My main job was to set up and
evaluate a high-availability virtualization cluster with 2 nodes based on
XEN and DRBD.
DevelerCompany's IT infrastructure - For over 6
years I've been continuously growing and restructuring
the network, the servers and their intricate mesh of
services and support scripts. The main server, called
trinity, contains 1TB of storage and offers file storage,
user authentication, e-mail, and many web-based services to a
highly heterogeneous and complex environment comprising several
versions of Linux, MacOsX and Windows clients. Additional servers
act as VoIP PBX and secondary slave for many (but not all) of
Fieremostre (Milan) - A cluster of 4 RedHat
servers, two web front-ends and Java appservers and two
database and filesystem back-ends with SCSI RAID5.
Fully managed remotely, including power fencing and
robotized tape juggler.
In 2003, I configured and installed a cluster of 5 nodes on
blade CPU boards with shared fiber-channel RAID storage.
I used a pre-release of RedHat Advanced Server
to implement a shared storage pool of 1.5TB with GFS1.
The system initially went to production without the GFS
pool because of reliability concerns with this new
Genexpress Lab - A combo of two interconnected servers, each
acting as a gateway and file server for a security ring. This design
was done for the Department of Bioengineering of the University of
Firenze, in Prato's Scientific Center.
My philosophy is using mainstream hardware whenever possible and
concentrate most services on few physical systems. The savings in
cost can be used to increase availability and performance through
redundancy. This strategy leads to data-centers that are easy to
understand and maintain, while at the same time scaling up much
better than traditional asymmetrical solutions (like web server,
mail server, db server...).
Being particularly fond of the UNIX culture, I tend to keep
my systems as open as possible while at the same time very secure.
TODO: add a list of server software I use/know
generated in 0.040963s
by GeekiGeeki version 4.0- 4cb7b4