Dave AitelThe #Neo4j Query Planner is honestly one of the best parts of it. You just put "EXPLAIN" in front of your query and then it shows you a graph with estimated cost at each step.
Bill
- 1 Followers
- 16 Following
- 4 Posts
@Pwnallthethings It's a genius plan to save on the bandwidth charges. If your servers are down, there's no traffic.
@Toxic_Flange The largest machines we had in the company were the E6900's. I never got to play with them though, the biggest systems we had in security were the 4500's we used for squid proxies. I will never forget the panic I had when I rebooted one and it appeared to be stuck after the prom messages. I don't remember why it needed a reboot now, but I do remember it was in the middle of the day. Called up our unix guys to help troubleshoot, at that point having waited five minutes on the most recent boot attempt. They laughed at me and told me to call back if it wasn't up in another five minutes.
@Toxic_Flange that keyboard reminds me of my first real career job building custom ipfilter firewalls on Sun UltraSparc 5's. I do not miss Solaris one bit, but the hardware will always hold a spot in my heart. Learned a ton about hardware while trying to make their bigger brothers, the 220Rs (might have been 280Rs) handling our DMZ traffic, pass a gigabit. I think we even tried a 4500 at one point before learning about the limitations of the 32-bit PCI bus.
Then came the surprisingly easy porting effort of moving all our code from Solaris to OpenBSD running on Intel hardware. Went from barely passing 200mbit to easily hitting 600mbit on boxes less than a quarter the cost. The gig goal was never really achieved, but the upgrade gave us plenty of headroom for production traffic. And a slight network redesign moved all the backup traffic which had been compounding the problem.