"Pageviews per kilowatt-hour (kWh)"Noted by Jim Buckmaster (Craigslist), they keep getting kicked out of colocation sites. I've heard this a lot lately, the biggest headache at highly scaled sites is the power consumption of all the server hardware. Perhaps it's simply a side-effect of the horizontal scaling (scale-out): lots of cheaper machines, rather than some big iron (but the latter is not affordable, and might not be suitable for this kind of job anyway).

So the challenge lies with chip manufactureres (CPU, RAM) and hardware companies: provide high performance with lower power consumption. One can probably simply look at 64-bit CPUs there, with the memory addressing capabilities and relative low cost (AMD Opterons being ahead right now, but Intel is coming out with new stuff), 32-bit for this type of server won't make much sense any more. Of course this is when talking …

This morning's elevator talk snippet (at OSCON)"So, which database do you use?"
"Yes."Turns out this company uses pretty much (each where appropriate) all - except Sybase, for whatever reason...
It was someone from whitepages.com.

Right now I'm in Tim O'Reilly's executive briefing which is a new separate event at OSCON. The first panel is on the impact of open source on web 2.0. Guests are Jim Buckmaster (CEO of Craigslist), Chris DiBona (Google), Jeremy Zawodny (Yahoo!). During their intro, they were actually arguing over who used MySQL the most ;-)
Jeremy reckoned it was Yahoo!, but you never know with Google. Craigslist is interesting in another way also. When looking at big succesful companies, most have thousands …

Thank you for the replies to my previous post.

However, let me rephrase the question:

I have a table of type MyISAM that is reporting 47 million rows when I do a SELECT COUNT(*). When I convert this table to InnoDB, running a SELECT COUNT(*) returns only 19 million rows. The conversion confirms 19 million rows were inserted and reports no warnings or duplicates.

I have done the conversion to InnoDB using the following ways
1. by dumping all the data in a text file and loading it.
2. by using ALTER TABLE

Why is the record count so low after conversion to InnoDB?
Who should I believe: InnoDB or MyISAM?
Any ideas as to what can be done to avoid loss of this many rows?

I will be posting output from my latest conversion attempt in some time.

----------------------------
PS: Thank you for all the replies. It turned out that the …

My Tutorial went well on the first day of OSCON. I ended up spending about 45 minutes on partitioning, and another 20 minutes on Events.

The surprise? I spent nearly 20 minutes on mysqlslap, our load generation and benchmarking tool in 5.1. I've run across a few system integrators who have been using it to "break in" new machines by generating SQL loads with many concurrent users, but this was the first time I have gotten a lot of user oriented questions about the tool.

For those of you in Portland this week MySQL will be throwing a reception/party on Tuesday (aka Tonight) at the Double Tree. Free drinks and t-shirt give away.

Free as in libre, since I don't believe we have in sales people here this week either :)

UPDATE I should also mention that the party is at 5:30->7:30 PM.

I watched the Falcon presentation from the MySQL Meetup in Boston. There was one optimization Jim spoke of that I was particularly fond of;

Falcon has a kind of two stage index retrieval. It first builds a list of what records it has to get, then buffers that and flips them into the correct order to scan through on disk. Potentially this means much faster retrievals; based on the assumption people are using one disk, or one of the RAID levels that doesn't support striping. Disk seek time is a real killer on performance in any database application, but I'm trying to think to myself now at what point the cost of software-sorting becomes more expensive, and this is actually a hindrance (since the raid controller will send the requests to different volumes anyway).

I'm guessing since disks are always a bottleneck, anytime you …

Well today is Monday, so this must be OSCON.  Actually, it's just been pre-conference tutorials so far, but MySQL was well represented with tutorials by Brian Aker and Jay Pipes on MySQL 5.1 in Depth and Maximum Velocity MySQL respectively.  And a group of us had dinner at a local Greek restaurant, which was fun.  Since MySQL employees are quite distributed, it's nice to get a group together at a conference.  We have speakers here at OSCON from Seattle, Australia, …

Just a quick reminder about the MySQL reception at the DoubleTree hotel tomorrow night (Tuesday) from 5:30 to 7:30pm at the Cantina Bar. We'll be giving away the famous uptime t-shirts, along with a variety of books, plus you get a great chance to meet MySQL developers and team members. Oh, did I mention the free drinks? See you there!

On another OSCON note, "Maximum Velocity MySQL" was a great time today. The only snag was a complete power outage about ten minutes from the end of the tutorial, but it was very well timed, as at the time I was going over the importance of having a battery-backed disk controller when using InnoDB. How very, very ironic, huh?

Last night, myself, Monty, Arjen, Brian, David, Julian, Zak, and Monty's son and daughter, Max and My, went out for Indian cuisine and had a good time. Tonight, some more MySQLers are doing the same; it might be the last time I get to have a good chat before the flood of main …

One week ago today I started work at OmniTI as a sales engineer for their Ecelerity MTA offering.

I’ve met and caught up with a great bunch of coworkers, learned a great deal about the new software I’ll be supporting, and generally had a really good time.

I think it is too early to really draw a comparison with working for MySQL, but it is a nice change of pace to sit in an office for a couple of weeks with the human interaction that goes along with it.

So what is Ecelerity? It’s a high-performance, extensible, clusterable MTA (mail transfer agent). Ecelerity is used by companies that need to send high volumes of mail such as Techtarget with their newsletters and reports. Where Sendmail may typically see 100K emails per hour of throughput, Ecelerity is marketed at one million messages per hour and sees several times that amount in the wild.

I’m in Maryland till Thursday then off to Seattle till Sunday, …