At Stanislav's request, here's my review of the test-month during which I had my blog up on a Rockchip he helpfully provided within his rack at Coloco Inc., USA. I did not take his offer to write this instead of paying for the month of service and I also preferred and insisted - again - to pay for this service1, as I think that any service used should be paid for in the end, one way or another. Or at the very least, I'd rather pay for what I use. Beyond that though, as he asked specifically for a review, I'm happy to do this for him.
My experience having my blog hosted on Stan's Rockchip was very smooth from a technical point of view: he transferred my blog from Pizarro promptly and totally painlessly for me, in the best way possible really - all I had to do was to switch the IPs and then start everything on the new machine. Some measurements of speeds from/to various locations I already provided in an earlier article and I won't repeat them. Throughout the whole month, there was only one blip - an instance when I couldn't access my blog - but so short-lived that I didn't even get the time to complain about it or indeed let Stanislav know at all. I have a knack of stepping in exactly at the right time to catch things and people on the wrong foot otherwise but I really don't have any complaints about this. For all the time otherwise, everything remained online and accessible and without any technical degradation of service that I could notice in any way.
Despite recent TMSR tribulations and the exposed rift in which No Such Labs fell to its death, I had initially all intentions of keeping my blog where it was, on Stan's Rockchip, as there wasn't any pressing need to move it, nor any specific problem with it being there as such, nor indeed any real trouble moving it at a later point whenever needed. It's a personal blog after all, not a business venture such as Minigame and as a result it can very well stay hosted in a friend's rack, why not. But then Stanislav complained in his chan of too many customers already:
asciilifeform: currently just under half the avail. wattage already spoken for. folx who are waiting, if wait long enuff, will find that train is full.
...and went on to add his quotes to my otherwise as-plain-as-possible words:
asciilifeform: diana_coman: 'no complaints' but 'not right thing'... aite, 'customer always right'(tm)(r) .
Your mileage might vary and I don't claim mine here to be objective at all - quite on the contrary, in this particular case for once, I'm letting it stand precisely at its most extremely subjective end and I choose for once to cut it out precisely on the line that works with *least effort* from me2. And so I choose to walk away for now and avoid dealing again with this sort of unexpected communications and not-business-but-not-quite-sure-if-really-friends-either service. As I previously said already publicly that I don't have otherwise any complaints from a technical point of view but I also don't have at all any certainty from an interaction & communications point of view, I'll simply preserve here the same words for easy access and completeness:
diana_coman: asciilifeform: I haven't forgotten about that, it's still somewhere on the list (though I'm not sure there is a lot *new* to say there); re service my main understanding as you pointed out is that you already have too many customers and moreover you anyway find me [http://logs.ossasepia.com/log/asciilifeform/2019-11-21#1002657][rather incomprehensible or a bothering-customer] so it is what it is.
ossabot: Logged on 2019-11-17 13:21:23 asciilifeform: currently just under half the avail. wattage already spoken for. folx who are waiting, if wait long enuff, will find that train is full.
ossabot: Logged on 2019-11-21 12:05:26 asciilifeform: diana_coman: 'no complaints' but 'not right thing'... aite, 'customer always right'(tm)(r) .
snsabot: Logged on 2019-11-20 19:47:57 diana_coman: http://logs.ossasepia.com/log/asciilifeform/2019-11-17#1002484 - I will make one space more empty on that train so that nobody else loses out; thank you asciilifeform for the service, I don't have any complaints with it really and I would still prefer to pay for this month that I used it.
ossabot: Logged on 2019-11-17 13:21:23 asciilifeform: currently just under half the avail. wattage already spoken for. folx who are waiting, if wait long enuff, will find that train is full.
diana_coman: if things change in the future and you have availability + looking for customers (me included), I'll happily use the service, yes.
asciilifeform: diana_coman: i in fact have 500 watt of capacity yet-unsold, fwiw. and diana_coman was a++ customer, is welcome to return any time, if she finds that the offerings meet her needs / at agreeable price.
asciilifeform: diana_coman: at the current rate (somewhat surprised there was interest) -- will likely buy the full tower w/ 20A & 1G/s in 2020.
diana_coman: asciilifeform: you know, from your communications and approach otherwise, I couldn't and can't really tell so the only thing I can reasonably do is to wait and see.
I sincerely wish that Stan's communications turn into reliable once - and when that happens, I'll gladly use the service. I also wish for him that the rack service truly takes off and morphs into a business that can give him otherwise the financial support and free time he is looking for.
I was willing to pay and even insisted on it right from the start; he agreed to receive the payment and invoice me only now, after I announced my decision to not subscribe to his service beyond this initial month. ↩
I earned my right here and in spades. ↩
Comments feed: RSS 2.0
Thank you for taking the time to write this review !
Would you be willing to elaborate re: "an instance when I couldn't access my blog" ? This is the first time I hear about it.
When was this, and for how long ? Did the box ping? (FWIW my logger lives right next to the box you were leasing, and has yet to experience a single unplanned disconnect. Could this have been an issue with your RK? Did it ever show sign of having reset? )
It wasn't reset for sure (for instance, I still had Apache on manual restart and there was no need for that). It honestly was such a short-lived blip that I basically didn't get to investigate it. It didn't ping either but that's about as far as I got so it was a matter of 15 minutes max (I had something else - as almost always - to do too).
If I knew anything further / had gotten there to anything, I'd have said earlier. Literally the weirdest blip ever and fwiw I do NOT hold it against your service and/or isp at all.
If this was a ISP-wide blip, it must have been strictly less than 3min. -- given as I did not lose my shell (to my logger RK, running since it was installed, 34+ days ago at the time of this writing.)
Still pretty interesting find (wish I knew about it when was happening though -- how to reproduce now?) -- and precisely why test pilotage is important.
Recall that you were "beta" tester, and installed prior to prices having been formalized. Which is why I did not invoice you until you explicitly asked. The box was specifically offered as a pilot demo, before I even had any data from which to judge whether the upstream ISP is worth a lick. We were giving *each other* service, of what I considered to be roughly equiv. value, and was pleasantly surprised that you considered it useful enough to pay for retroactively.
Re: communications of the life and times of my rack -- installed the 1st subscribed "dulap" last night, and expecting to install a second shortly. Negotiated with colo house re a possible upgrade of the mains supply, but result was a quote for an outrageous price (+250/mo) -- clearly intended to say "buy the full tower" (where same 20A in place of 10, but additionally 1G/s in place of 100M/s, for only +200/mo).
Will buy the tower, as said previously, only when the demand pressure justifies it. And indeed this is "ugly" from POV of folks who want "industrial" ISP, where can order e.g. 9000 boxes on a whim. However I knew what I signed up for -- explicitly wanted ISP that can remain "pocket-sized" if it must, such as to properly fill my own machine needs, and at reasonable expense, regardless of what other people may or may not wish to buy. I have no complaints at all with the buyers and the non-buyers.
There was no mention or intention of "order 9000 boxes on a whim", that's entirely your perception and idea. Please do not pass it as having *anything at all* to do with me.
Re asked and just for readers so it's clear: I asked for at least 3 times and it's all in the logs linked from the article above, it can be checked.
For the rest, I hope your service takes off/stays same exactly the way you prefer it really.
The "9000" wasn't meant as a criticism of your stated purchasing reqs, as such; but is a reference to a related and very real problem that I freely admit I do not know how to solve, or whether is even solvable as-stated. And it is a broader problem than what has been discussed earlier, and IMHO does not at all reduce to "shortage of capital."
It is not clear to me presently that "scalability" of the kind contemplated in the original conversations re "build an ISP" can be reconciled with the hygienic requirements which (cannot speak for others) I demand for my own irons, and in good conscience could offer to customers.
To run with the "chicken farm" analogy -- I won't sell to others chickens that I myself would not eat. Would happily grow and sell "free range" birds, all beaks still attached, fed 100% on earthworms, fed 0 antibiotics. But it is impractical to scale these up to where could even in principle compete with "factory farm", either on price or volume, i.e. where the work load extends beyond what any one pair of hands could do. There is a reason why (AFAIK) 100% of mainstream ISPs use e.g. BMC remote control, buy brand-new "Dell"s with NSA-blessed CPU, etc. Not even to mention that they do not limit their clientele to WoT.
If I knew how to reliably and quickly "make copies of self" -- could scale everything I do ad infinitum without compromising on quality. But I do not.
Stan, they are *not my purchasing requirements at all*, that is the thing! Anyways, now it's stated clearly so hopefully we can leave it at that.
To answer your raised issue re scalability there though, I think that *can* be solved namely via WoT simply. Yes, you can't clone yourself and yes you can't service yourself just any number of servers. However, you *can* train your *own* people and therefore scale it people-wise, not machine-based. As those things are always done really, there aren't that many options.
Sure, I hear you if you say that no, can't find nor train nor trust nor whatever other people and in that case it remains where it is and entirely on your own back for as long as you can still carry it along all the other things you make-to-carry, there's nothing anyone can do about it anyway.
Experiments with growing and training armies of people who "can ISP" are probably "where it's at", in long term, I'll have to agree. But given as such hands have to physically live on-site (as e.g. BingoBoingo did) they will have to happen somewhere outside of USA and similar ludicrously overpriced real estate -- while at the same time somehow accessing proper backbone-grade telecom.
In USA in particular, racks may be cheap, but humans are almost prohibitively expensive. And, in my experience, ones fit for any such training -- quite uncommon, and virtually never seen on labour market.
So indeed I "built what can be carried on own back".
I can't help but notice - and to some initial surprise really - that ~everyone in #o is actually from the US. Sure, not on the labour-cattle-market, no surprises there.
Anyways, there's always but aplenty, yes.