5 ms·

well by asking this question to mongo users you miss the audience of the people successfully scaling up -for less money- that may belongs to the people NOT usin

well by asking this question to mongo users you miss the audience of the people successfully scaling up -for less money- that may belongs to the people NOT using mongo. Business is about costs and prices not about thinking that business is about the religious adoptions of mongo, agile, C#, python, linux .... This question is saying that your business is ruled by the techs not anyone with the common sense needed to pay the bills (thus your paychecks). The feedbacks I have to give you is that TCO per mb stored increase more than linearly with mongodb, with weiredly enough a decrease in SLA (due to the weired unexpected incident hard to troubleshoot) so if you look at the probability you have success, it is bad news. Customer expect cost to diminish with volume, thus prices (investors too). If the fat of your business model is around document storage, you have a problem. You should not trust me, and do some real costs analysis of the cost per documents for your business case with more than one scenarii and make some costs analysis and some projections. Because even if someone else optimization worked for them, it may not be relevant to your business case: at one point you make money, so you issue a transaction that cannot be handled in a distributed system (CAP theorem + NoSQL==NoACID) and this point differs than fellow devs. Mongo could be -not is- killing your business. If you do mongo because that is the only techno you know, consider that you have been taken hostage, and call SWAT to have your mongo guru arrested. They may look nice and harmless, but you are the hostages of dangerous fanatic religious techs. Maybe call an exorcist, it could be as efficient. And then go back to thinking what is the most efficient way to have your bills paid, and how to diminish cost as your business grow. Forget technical stuff as a compass for making successful software solution. Success is about benefits only Je dis ça, je dis rien, hein?! My .02 eurocents.

mikegioia 12y ago
This is an unbelievable waste of a comment and the OP should not even consider this as feedback. You completely ignore the question, make a comment about "religious adoptions of mongo" and then do an about-face by showing your OWN religious zeal AGAINST mongo. Mongo is not killing his business. Your comment should be deleted. To the OPs question: can you post any more specifics? 1M documents with an index shouldn't be an issue for the aggregation pipeline. What are the specs of the machine? What are your indexes? I think it would be tough to help without having that info, and sadly a question requiring that type of info may be best suited for something like Stack Overflow.
- Iftheshoefits 12y ago
  I don't know, it may not address the OP's specific question but it may, in a very circuitous way, address a need the OP doesn't recognize he has. It is hardly arguable that there is a segment of the developer population that has a religious zeal for their "stack" or against others (just pick any random topic on functional programming or PHP for examples of each). It is entirely appropriate to discuss and address that in topics like this on HN, in my opinion.
  - baq 12y ago
    this religious thing works the other way, too. e.g. i, after reading a couple dozen articles about mongo across the years, most of them linked from this site, have a religious zeal against mongo.
    - MichaelGG 12y ago
      A database that had a server-wide lock (and highly trumpeted when they got the feature of only being database-wide) and had the amazingly in-depth IO strategy of mmap, defaulted to losing people's data, silently destroyed data on 32-bit, had a Java client that shipped with "catch { if rnd() > 0.1 then .... }", and spends a lot on money trying to be "webscale"? Perhaps it's a deserved attitude.
    - Iftheshoefits 12y ago
      That may well be the case. HN has a community of zealots for just about any programming or tech related thing (pro-Apple, anti-PHP, anti-C++, pro-functional, pro-javascript, you name it). That doesn't mean criticism (or support) automatically comes from a place of zealotry. In this particular case the parent comment in this thread didn't read at all like religious zealotry against mongoDB to me. I suppose one might infer from it notion of a default assumption that developers are all zealots about their particular technology choices. I'm not sure such an assumption itself is terribly bad to have, honestly. Sometimes I think that if my only insight into the world of technology came from startup job postings and HN, I ought to be forgiven for thinking that the entire industry is a kind of elaborate and humorous play on role-playing game mechanics ("Ruby Dev" <==> "Developer/Ruby" class/skill combo), which are notorious for producing fanatical devotees of one combination or another.
danpalmer 12y ago
This is an interesting way of looking at the problem - as a 'TCO' of data stored. It's not one that I generally go to, but perhaps I should more. I think the argument against it is that developer time and productivity is worth a considerable amount for many companies, and working with tools (DBMSs in this case, but it applies to lots more) that speed up application development is often worth more to the business than the basic cost of storing and serving data. This isn't always the case, at Facebook scale for example, it might be that the most efficient use of money is to use (speaking entirely hypothetically here) is an Oracle database instead of MySQL, and instead hire twice as many developers because they are half as productive with it. Maybe that would be a good trade-off? However, at many (most?) companies, I suspect that the better trade-off would be to hire fewer developers and use tools that are more expensive to run.
- themonk 12y ago
  And after that pay big support fees to Oracle.
- Spooky23 12y ago
  There's the risk tolerance factor as well. If product X is free, but requires hard to find skilled operators and doesn't have well established operational best practices, paying über dollars for Oracle might be the smart move. Where I work, location and policies for hiring contract staff make a product like Mongo high risk for data we care about.
- mikegioia 12y ago
  You're being trolled by SFJulie1.

mikegioia 12y ago

This is an unbelievable waste of a comment and the OP should not even consider this as feedback. You completely ignore the question, make a comment about "religious adoptions of mongo" and then do an about-face by showing your OWN religious zeal AGAINST mongo. Mongo is not killing his business. Your comment should be deleted. To the OPs question: can you post any more specifics? 1M documents with an index shouldn't be an issue for the aggregation pipeline. What are the specs of the machine? What are your indexes? I think it would be tough to help without having that info, and sadly a question requiring that type of info may be best suited for something like Stack Overflow.

Iftheshoefits 12y ago
I don't know, it may not address the OP's specific question but it may, in a very circuitous way, address a need the OP doesn't recognize he has. It is hardly arguable that there is a segment of the developer population that has a religious zeal for their "stack" or against others (just pick any random topic on functional programming or PHP for examples of each). It is entirely appropriate to discuss and address that in topics like this on HN, in my opinion.
- baq 12y ago
  this religious thing works the other way, too. e.g. i, after reading a couple dozen articles about mongo across the years, most of them linked from this site, have a religious zeal against mongo.
  - MichaelGG 12y ago
    A database that had a server-wide lock (and highly trumpeted when they got the feature of only being database-wide) and had the amazingly in-depth IO strategy of mmap, defaulted to losing people's data, silently destroyed data on 32-bit, had a Java client that shipped with "catch { if rnd() > 0.1 then .... }", and spends a lot on money trying to be "webscale"? Perhaps it's a deserved attitude.
  - Iftheshoefits 12y ago
    That may well be the case. HN has a community of zealots for just about any programming or tech related thing (pro-Apple, anti-PHP, anti-C++, pro-functional, pro-javascript, you name it). That doesn't mean criticism (or support) automatically comes from a place of zealotry. In this particular case the parent comment in this thread didn't read at all like religious zealotry against mongoDB to me. I suppose one might infer from it notion of a default assumption that developers are all zealots about their particular technology choices. I'm not sure such an assumption itself is terribly bad to have, honestly. Sometimes I think that if my only insight into the world of technology came from startup job postings and HN, I ought to be forgiven for thinking that the entire industry is a kind of elaborate and humorous play on role-playing game mechanics ("Ruby Dev" <==> "Developer/Ruby" class/skill combo), which are notorious for producing fanatical devotees of one combination or another.

danpalmer 12y ago

This is an interesting way of looking at the problem - as a 'TCO' of data stored. It's not one that I generally go to, but perhaps I should more. I think the argument against it is that developer time and productivity is worth a considerable amount for many companies, and working with tools (DBMSs in this case, but it applies to lots more) that speed up application development is often worth more to the business than the basic cost of storing and serving data. This isn't always the case, at Facebook scale for example, it might be that the most efficient use of money is to use (speaking entirely hypothetically here) is an Oracle database instead of MySQL, and instead hire twice as many developers because they are half as productive with it. Maybe that would be a good trade-off? However, at many (most?) companies, I suspect that the better trade-off would be to hire fewer developers and use tools that are more expensive to run.

themonk 12y ago
And after that pay big support fees to Oracle.
Spooky23 12y ago
There's the risk tolerance factor as well. If product X is free, but requires hard to find skilled operators and doesn't have well established operational best practices, paying über dollars for Oracle might be the smart move. Where I work, location and policies for hiring contract staff make a product like Mongo high risk for data we care about.
mikegioia 12y ago
You're being trolled by SFJulie1.