3 ms·

Top Internet Startup Scalability Killers

15 points by hshah 17y ago

kevinholesh 17y ago
In my mind, scalability is the best problem a startup could ever have. It means people demand your product so much that you need to concentrate on making more and more of your product. Definitely better than too little demand.
- strlen 17y ago
  It depends on what you do. Scalability is about building a system to handle ten fold more traffic than it's seeing now; <10 is too little, 100 is over-engineering (and nearly impossible). Now imagine a graph data structure (any site with a news-feed/following functionality). You can get 10x increase in edges without a 10x increase vertices. Some kind of products also increase in usefulness (and demand) as they increase in user growth, so solving scalability problems is necessary to their success. It isn't, however, sufficient.
gfodor 17y ago
You could just prune away everything except these: - Retain competent people - Fire incompetent people - Get out of their way
- rbranson 17y ago
  Totally agree. I don't think anyone would have looked at LAMP 5 years ago and say it's scalable, but Facebook did it, in steps, because they had very talented engineers. I found the point of "don't use a relational database" ironic considering they contrasted Friendster and Facebook. Facebook uses MySQL as their backing store. While they might not use it in the traditional way, they most certainly evolved from joins and normalization to their current scheme as a sharded object store as it became critical to their growth.
  - ams6110 17y ago
    The point was not "don't use a database"; the problem identified was "inappropriate use of a database." And that is a problem which is much larger in scope than the simplistic examples discussed in this piece.
sokoloff 17y ago
Seemingly far more people are concerned with the scalability of their startup than the viability of it. I'm coming up blank with a startup that would have been successful but was killed off by scaling issues. (Twitter may have come the closest...)

15 points by hshah 17y ago

In my mind, scalability is the best problem a startup could ever have. It means people demand your product so much that you need to concentrate on making more and more of your product. Definitely better than too little demand.

strlen 17y ago
It depends on what you do. Scalability is about building a system to handle ten fold more traffic than it's seeing now; <10 is too little, 100 is over-engineering (and nearly impossible). Now imagine a graph data structure (any site with a news-feed/following functionality). You can get 10x increase in edges without a 10x increase vertices. Some kind of products also increase in usefulness (and demand) as they increase in user growth, so solving scalability problems is necessary to their success. It isn't, however, sufficient.

gfodor 17y ago

You could just prune away everything except these: - Retain competent people - Fire incompetent people - Get out of their way

rbranson 17y ago
Totally agree. I don't think anyone would have looked at LAMP 5 years ago and say it's scalable, but Facebook did it, in steps, because they had very talented engineers. I found the point of "don't use a relational database" ironic considering they contrasted Friendster and Facebook. Facebook uses MySQL as their backing store. While they might not use it in the traditional way, they most certainly evolved from joins and normalization to their current scheme as a sharded object store as it became critical to their growth.
- ams6110 17y ago
  The point was not "don't use a database"; the problem identified was "inappropriate use of a database." And that is a problem which is much larger in scope than the simplistic examples discussed in this piece.

sokoloff 17y ago

Seemingly far more people are concerned with the scalability of their startup than the viability of it. I'm coming up blank with a startup that would have been successful but was killed off by scaling issues. (Twitter may have come the closest...)