4 ms·
Maybe you should read the other posts about future plans and how this is extremely alpha. Or maybe go anywhere else and do literally anything else but be obnoxi
by alangibson 5y ago
Maybe you should read the other posts about future plans and how this is extremely alpha. Or maybe go anywhere else and do literally anything else but be obnoxious in this thread.
I'm spending over $200 per week just to stand up the service as it is. $10 for a full functional search engine will likely not be even close to PROFIT!!!!
- sdesol 5y ago> I'm spending over $200 per week just to stand up the service as it is What are using if you don't mind me asking? Not trying to criticize or anything. I have a Heztner box that gives me 1TB SSD in RAID 0 mode and 64 GB of RAM for about 80 CAD a month.
- alangibson 5y agoI have 3 sizeable EC2 instances running an ElasticSearch cluster, plus a beefy box for data preprocessing and crawling. A big chunk actually goes to BigQuery. There are publicly available datasets for HN, Stackoverflow and a few others there. I've also loaded up the Common Crawl index. The query and storage fees really add up. I'm hopefully done with huge BigQuery queries, so that $200 will probably drop for a while.
- rmbyrro 5y agoI'm probably wrong about my assumptions, but presume you are open to any kind of constructive feedback, so here it goes... Maybe you're overkilling with the infra stack. I would simplify until having a mature product, especially if I'm bootstrapping, which I think is your case. Right now, you're still a bit far from MVP, from my point of view. Those $200 can probably be reduced by 50%-75% if you compromise on stuff only important to non-alpha services (i.e. 99.99% availability). A single EC2 box should be enough. Maybe look into Postgres or another FOSS instead of BigQuery. These $100-$150 savings per week can go into promoting your service, getting as much attention as possible to maximize feedback. Good luck!