Sunday 29th of January 2012

on Sunday, January 29th 2012 I 0 Comment

 Sunday 29th of January 2012
Last week, we were switching on our new file server architecture (NAS) for 2012.

The incident of October made us buy lots of material:

- 40
SEAGATE ES disks (Enterprise Class) guaranteed for 5 years and 1.2 million hours. 2 To each fo the NAS.
- 20 Intel SSD Drives 600 GB at 1000 € each,
5 year guarantee, in order to have an ultra fast pool.
- 2 HP servers used as NAS head with 16 core CPU, 16GB of RAM and dual 10Gb network interface.
- A P2000 24 drives HP bay with quad 10 Gb network interface.


We divided the 20 SSD between Sandy 1 (NAS 1) and Sandy 2 (NAS 2), they are 100x faster than hard-drives with trays. We changed all disks in Sandy 2 by 20 Seagate ES disks  2 To


We have put in production a pool of 10 disks on Sandy 1, we have earned an average of 30% in speed. We also launched the synchronizations of data on Sandy 2.

During the whole weekend we monitored the synchronization in order to check the transfer of data and prevent slowing down all sites. All should have been finished on Monday as Sandy 2 should have been put into production in the afternoon.


But today, Sunday at 4:00 pm, 4 new Intel 600 Go SSD were simultaneously lost. The system can support a loss of two disks at a time but not 4. We had a back up J-1. We thought we could still use the snapshot made on Saturday but the data were not usable. For articles published after Friday, January 27 after 1.55pm, you will have to upload the photos and attachments again, however no worries on the videos.

In our misfortune we've had some luck because two disks were back working for a few minutes. We immediately made ​​a snapshot of the data which allowed us to recover the differential data.

HP and Intel

We ave been very cautious since October, we have mobilized several people full time on this project. But we did not count on of HP and INTEL letting us down!


HP took a month and a half to tell us it was normal to have a bandwith of 700 Mb/s while the technical document of the bay announces 1600 Mb/s. And I am not talking about the servers for which we took one month to find out a bug in the system energy savings that was rebooting the servers during the night. If HP wants information about it, we can  help them out because their servers are in efficiency mode when the load is too low. What is outrageous is to buy 100 000
€  of equipment and their support is worth of the one of a washing machine.


We bought 20 disks at 1000 € / item at Intel. They post an alert on their disk saying that we must update their Fireware otherwise we risk a Bug. INTEL is (was) the best brand of SSD. OVH confirmed that they only use this brand for reasons of reliability. These disk are 30% more expensive by Go. Well, late yesterday afternoon, despite the supposed Fireware updated fixing the bug and the 5-year warranty, 3 discks have lost their data.

In conclusion:

I'm not trying to put the blame on HP and Intel, but honesltly, they are not second-tier manufacturers and they make us pay dearly for their brand. We've put on a very temporary WMaker Pool SSD that we are going to cut off as soon as the back up on the hard drive will be finished (end of day Monday).

Monday we will ask a quotation at NetApp which has the reputation of being the Rolls Royce of NAS. We will probably have go through them for our peace of mind.

This year is starting well + 1.6% VAT and minimum 60 000 €  for 2 additional NetApp NAS ....

We will not give up, we have lots of great news to be published in February, we will carry on!

Saveria Maroselli
As support manager, Saveria guides you daily while you're using your favorite CMS, but not only,... Know more about this author

New comment:
Facebook Twitter