What can be done so that the tricks of any Makhost are not scary?

    The prelude is known to all. Makhost behaved badly, Oversan, Space Technologies and Hosting Community rubbed and bargained for a long time, but now everything works. I want to believe that it will continue to work, and maybe even without changing ip-addresses, prices and other conditions.

    And if not? One hoster, one DC, one colocation site still have an insignificant, but the probability of a fall. What to do?


    It is clear that they must be fresh, affordable and full. And they should be. It is amazing how many people have them old or "but I did not merge the database."
    I have only 2 questions:
    1 . I have backups. These are 600 tar.gz files located either in Ireland, or at Agave, or at the site of the partner with whom we exchange backups. How long does it take to get 300 working sites out of them with mailboxes, crowns, parsers configured and all kinds of data imports? And also the panel on which everything is tied.
    2. Maybe I’m a teapot and I don’t understand anything in sausage scraps, but removing a full backup from a running server causes me slight doubt. As we pour, the data changes. Files, databases, letters come and go. At a minimum, this is an imperfect, incomplete copy. At most - inoperative.
    And, it seems, no protocol eliminates this. Comfort me?

    Spare area for key customers

    We have an online store 700shin.ru (now NSs are changing back, you won’t see it). Crowns, orders, two-way synchronization, etc. Mail from Google.
    For a long time already, in order to avoid such ambushes, a synchronous mirror was made on 1gb.ru - http://alt700shin.ru/ . There, if necessary, the main domain is also redeployed.
    But fuss with him a lot. The solution is expensive in terms of money, in time and not universal.
    Mirroring issues are complex and not fully understood by me, but this seems to be the solution.

    Belief in foreign hosters

    Maybe there is better. But that there was 100% reliable - I do not think. Problems in the event of accidents or fraud you will have a larger caliber than in this country.

    Server under the table

    Many do this, now even more will be those who want to. But the problems are known and not very pleasant: a thick channel, pings, backup electricity, the Internet. Even if you are of the same qualifications as the hosting staff, you will not solve all the problems.

    Any other solutions? I did not come up.

    I came up with this. It may be expensive and at first hemorrhagic, but in case of accidents it can save.

    Solution option

    I have a good server in the host, anlim on traffic and raids.
    A server with the same screw is bought, a mirror with the same OS, panel (license issue is being resolved) and data is poured onto it. Let there be a weak percentage and an expensive channel, it is important that this is all workable in case of an accident. Let it stand in the states.
    The daily (weekly if we do not pull in speed) synchronization of everything is configured. Panels, mails, databases, files, logs. As far as I understand, this will not cause problems for Debian and other niks systems.
    Since all my 300 domains have 1 and the same NS, which we also support, in case of an accident it will be necessary to specify a new IP for this domain. Maybe this can be solved both through automation and faster, without redelegation.

    Then, in the case of major or minor troubles, you and the beer and popcorn watch the shit boiling over, but you are not at all as scared as the others.

    Questions: who to choose, how to set up synchronization, how to switch quickly and how much it will cost. What problems have I not noticed?

    We pay Makhost for all 9200 a month. If such a backup solution will cost 4600 (another piece of iron and another channel), I agree to seriously consider it.

    Also popular now: