OVH Community, your new community space.

Dedicated Servers: nice refresh

02-02-2014, 00:19
A little refresh of the dedicated servers range in version "nice -n 15", which can be summarised as described below. We are announcing that the entire 2014 range of OVH dedicated servers has now been launched - Wahey!
When can we expect brokers website?

30-01-2014, 12:30

A little refresh of the dedicated servers range in version "nice -n 15", which can be summarised as described below. We are announcing that the entire 2014 range of OVH dedicated servers has now been launched - Wahey!

Here are the details of our "nice -n 15":

- EG-128G
A new server is available in the Infrastructure range. Equipped with 1 x E5-2650v2 8c/16t with 2.6 GHz, 128 GB RAM ECC and 2 x 10G, it's the perfect server for setting up web proxies between the public network and the private network in vRack 1.5. It costs €190 excl. VAT/month OVH uses it for the CDN, GeoCache, pCS and soon pCI, vRack, and hubiC proxies, etc.

- HG
The standard CPU on the HG servers changes from 2 x E5-2630v2 12c/24t 2.6 GHz to 2 x E5-2650v2 16c/32t 2.6 GHz. Without any increase in price. The price of CPU options comes down.

The standard CPU on the BIG HG servers changes from 2 x E5-2650v2 16c/32t 2.6 GHz to 2 x E5-2670v2 20c/40t 2.5 GHz. Without any increase in price. The price of CPU options also comes down.

Want to know how this is possible? Read the "Why" bit at the end of the message. In all cases, customers that already have a 2014 HG or BIG HIG server will be able to benefit from this upgrade:
- change of CPU to the new one by default
- reduction in price for the CPU option in OPEX mode only

- The FS series
We are setting up a series of 6 hard drives of 12 TB to... 216 TB per chassis!! We have specially designed these servers for storage with the integrated vRack 1.5. We use it for our pCS requirements, and conseqently, for hubiC.

Using these servers you can create:
- fast storage
- slow storage
- OpenStack Swift
- Ceph
- backups
- multi-datacentres

The FS series also integrates a new type of HGST Helium Enterprise disk of 6 TB capacity (!!) which can therefore provide 72 TB in a 2U and 216 TB in a 4U.

The servers with 36 disks are connected on 2 LSI HBAs cards of 48 Gbps each (8 channels of 6 Gbps SATA3/SAS2). So inspite of the large number of the disks in the frame, we can guarantee that the maximum speed per disk is limited by the disk capacities, and not by the LSI card.

You get the the Hybrid option of a mixed SSD/HDD, making the servers compatible with the Ceph specifications. The server is compatible with the Openstack Swift specs by default.

The FS-72T and FS-218T servers will be available by mid-February, we are one of first suppliers to receive these new 6 TB Helium disks, which are not yet available in large quantities. For the time being, we are waiting for these little gems to arrive in our hands.

The FS-144T and FS-218T server will integrate a brand new feature (!!) which should be approved in the 10 next days. We cannot tell you more than that, but it will be worth waiting a little longer…

The other FS servers are available, and we will work on making them available in our 4 datacentres zones: BHS, GRA, RBX and SBG. For example, hubiC has been synchronised on the 3 DCs GRA, RBX and SBG.

- The HPC range
We bring our Oxalya subsidiary company to the forefront, which specialises in HPC and the cloud solution that we have developed for parallel computing.

Our software solution enables:
- uploading of models to be calculated, for example, the flow of liquid in a tube
- launching calculations of the model to see how the fluid and the tube react to pressure and temperature, etc.
- remote visualisation (!!) of the calculation results, to select the most interesting aspects
- retrieval of the calculation results

We are the only provider on the market with a functioning solution that enables these 4 elements to be carried out. In particular, we have a image flow compression technology which enables the results to be viewed with very little flow, and which has the advantage of functioning and or even functioning perfectly, with HP HPC graphics cards for example.

We are offering 3 ranges of service:
- a shared HPC cluster (hourly or monthly)
- a dedicated HPC cluster (hourly or monthly)
- a HPC cluster installed on your premises (Oxalya's historical activity) pure CAPEX, with our software solution as an option. We are a integrator of HP servers for the HPC industry. We will soon be offering hourly or monthly mono-servers)

- The Big Data series
4 servers designed for big data. The servers have JBOD disks with perfectly balanced CPU/RAM/HDD ratios for all big data purposes. The servers will be available for order from mid-February onwards, as we are waiting for delivery of an essential element required to make authentic big data servers. We will announce it as soon as we receive it. The BD-48T is also awaiting approval of a new feature. The same as on the FS-144T. We can start guessing what it might be…
For 3 years, we have been offering AMD 4 x CPU. With the Big Data range, we have also approved our first Intel 4 x CPU: 4 x E5-4607v2 24c/48t with 2.6 GHz with 512 GB RAM and 24 x 2TB. It won't be the last.

Why were we able to upgrade the HG and BIG HG servers without increasing the price?
As we already announced, in order to supply you servers with the fastest deliveries, we have reviewed the relationships with all our suppliers. Basically, we supply our assembly lines with components that come from all corners of the world, and it's important to have everything at the same time in order to be able to manufacture a server. Building more and more servers, which are all different from one another, with supply deadlines of up to 3 months... it has become ultra complex. It was necessary to evolve and to re-examine things, and we have done it.

So we became an OEM customer, which has enabled us to create the supply lines for components with 24/48hr delivery times. What's more, we have also committed ourselves to a volume of components to be purchased throughout the year. There's really no risk as we would be buying it anyway.

This enables us improve how we manage our stock, and to be more reactive to the success that some strategies may bring. Got a big order to place? No problem, within 48hrs we will have the parts and we can deal with high order volumes without having to buy a huge quantities of parts in advance. It's also essential to supplying almost 180 different models of servers!

Signing these contracts has enabled us to reduce the price of some components. Cool. But instead of just looking to reduce our costs and increase our margins, we have taken the opportunity to improve the deals we are offering you. Even better for the same price. HG and BIG HG are concrete examples of this strategy. Also, The FS and Big Data ranges benefit immediately from very competitive prices.

We hope to offer you the best possible hardware at the best price, while continuing to innovate. It may also happen that other server ranges may benefit from some improvements in the coming months. A few tweaks here and there, rather than big announcements and big changes. With 180 models, we simply can't update the whole range in 24 hours.

Thanks to all these preparations, the availability of OVH servers is now being managed correctly - we deliver in 120 seconds, and in 2/3 hours for customised servers.

We are currently working on the availability of all OVH servers in all DCs. As we won't have any more space in RBX before 2015, (except for EG/HG/BIG HG/pCC, which we reserved in advance) we have to have set up room "fragmentation" projects... oh the joys of managing a total of 165,000 servers in nearly 5000 racks, hosted in 4 zones! We're also working on the availability of new range that we have just launched. So far so good.

Now that the OVH ranges have been launched, we can keep ourselves busy with SYS…