Moving is always stressful, no matter what it is. Moving from a less comfortable two-room apartment to a more comfortable one, moving from city to city, or even pulling yourself together and moving out of your mother’s place at 40. With the transfer of infrastructure, everything is not so simple either. It’s one thing when you have a small site with a couple of thousand hits per day, and you’re willing to spend several hours and a couple of cups of coffee transferring data. Another thing is when you have a complex infrastructure with a bunch of dependencies and crutches placed in certain places in a specific cloud.
And if you add 1C to this, then the process begins to play with new colors.
My name is Sergey Kondratyev, I am responsible for our striped cloud, BeeCLOUD, and in this post I will tell you about the move of the AeroGeo company to our cloud.
Why move at all?
First of all, let’s talk about the specifics of AeroGeo’s business. This is a Krasnoyarsk airline that has been transporting passengers and cargo for 13 years; they have more than 40 aircraft in their fleet, including helicopters. They fly only in Russia, but throughout the entire territory. That is, the company’s aircraft can be found from Altai to Kamchatka. The fact that AeroGeo ensures the full operation of the Seasonal Drifting Station of the Russian Geographical Society has become a kind of calling card.
Bell 429, photo from company
In general, there are enough clients, more than 350 internal employees, aviation work of any complexity. Therefore, an adequately functioning infrastructure for a company is very, very critical. And you know how capricious 1C systems can be even without me.
So. A year ago, the client had a clear need to update the infrastructure. Of course, they began to look towards working cloud solutions, and then it turned out that, firstly, the company’s management had a little doubt about cloud solutions (whether everything would actually be available 24/7 or not), and secondly, They definitely didn’t want to work through a public channel. We must give them their due; when we decided to move, they gave us a serious check: the IT director personally flew in to look around the place and understand what and how it works for us. I walked around, looked, drew conclusions and gave the go-ahead for the pilot project.
The infrastructure that needed to be transferred was designed for the work of 30 specialists at a peak from three different offices (read - from three different networks, the head office, Yemelyanovo airport and AeroGeo airport). We thought about it and decided to combine all this into a single network, which we then reserved using the IPSec protocol, and installed a dedicated 100 Mbit Krasnoyarsk-Moscow tunnel. The hardware key is located in our data center on a USB hub and is transferred to the client’s pool.
The migration took only one evening, because the representative of AeroGeo simply took and brought us the main database on physical media directly to the data center in which the platform was deployed. Actually, we were worried about key binding; there were a number of fears that the keys would fall off during migration, but no, everything went fine, because the keys were bound to similar hosts.
The pilot project lasted about a month, we actively collected feedback from 1C specialists. During this month, they did not notice any drops in productivity or inconveniences.
Why come to us
There are a lot of clouds now, almost every major player on the market already has its own cloud with a bunch of goodies. It’s understandable, if you want to compete, make a great cloud and a little more on top.
We currently have three data centers (Moscow), a cloud on OpenStack (if you are interested, I will write about this in detail in a separate post), we have managed to get our hands on transferring very different 1C systems to the cloud, BeeCLOUD has hosts at 3 GHz, and at 3,5 GHz (just the same, with a dedicated HP Synergy cluster at 3,5 GHz, was chosen at AeroGeo), depending on what the client requires.
And since 1C is such a thing that in setting it up and finishing it, the principle “Who cares” continues to actively operate, we made an excellent cluster where the client can drag his most customized, capricious and hardware-demanding 1C and not spill anything along the way. Everything will work. TIER 3, SLA 99,97, FZ-152, classic scenario.
But these are all numbers and technologies. Our product is all about people. We managed to assemble an excellent team of cool engineers who are based both in Moscow and work distributed in the regions. This gives us a very important opportunity to help the client locally. It’s one thing when you (even as a VIP client) call support and hang on the line for a while, explaining what’s broken this time, after which support comes to check everything remotely. It’s another matter when networkers and experts are able to solve all potential problems on the spot, with these hands.
Of course, the cloud is also good because it removes all the headaches from the client and solemnly transfers them to the provider. At AeroGeo, everything was tied to this 1C. Now they know that we keep the system up to date and operational. Something new comes out from the vendor, we need to roll out some kind of patch, etc. - we simply write to the client about it, agree on a convenient time in his time zone for the work to be carried out, and work. For example, when fresh patches from Intel and HP were rolled out to the hosts, this was done by our guys during the lowest load Krasnoyarsk time.
We also managed to do everything within one window. Different services sometimes have problems in that it seems like you, as a provider, provide a service, but you have a lot of contractors. And if something goes wrong with the contractors, then time is also wasted on communication with them. The client doesn’t care, since he pays you, then you should solve all the problems.
Therefore, in the case of BeeCLOUD, we decided to move away from this and do everything ourselves. Your own main channel, your own support, your own hardware. This is also faster for the client in case something happens, if some problem arises, it means that this is definitely our problem, we will solve it. Plus it greatly (in fact) saves time on internal processes when you have everything of your own - you have one service desk, without a bunch of clones and synchronizations or constant ping-pong between contractors.
And about money
Where would we be without this? I can’t reveal many numbers within the framework of this post, but they will still make it clear the scale. When AeroGeo calculated how much it would cost to modernize the existing infrastructure, they calculated more than 2 rubles. And this is preliminary data, the kind that usually comes from papers marked “From.” Update only, no maintenance or support.
For the infrastructure transferred to BeeCLOUD, including the capacity itself and round-the-clock support, the client pays 45 rubles per month. That is, two million rubles will be enough for almost 000 years of work without fuss and other things.
We try to be as open as possible, if a client wants to come to us and see how everything is going - please. By the way, about the cloud itself: .
If you have questions about this case or about our cloud in general, write to me, I’ll be happy to answer.
Source: habr.com