From the nuts and bolts to how you present data to your visitors - you can squeeze speed out in many ways.

The (relatively) popular talk is about optimizing the user-download sequence to wring out as much performance as you possibly can. I’ve talked about this in the context of CSS sprites (great for speed oomph).

But there are other things you can do at the server-level that often get skipped over.

Two quick examples:

Our weather widget is pretty popular. It is now used on over 200,000 unique pages. Every pageview on one those pages = a call to our server. So of course we cache the HTML we output, but that still means our hard disks get hit for every request. So we looked closely and saw that the cache never grew bigger than 2 gb. The solution? A server with 8 gb ram, 4 gb for normal memory options, and a 4 gb ramdisk that is used for the cache output. This saves a lot of wear and tear on the HDs themselves, while letting us respond faster than ever.

Second example - re-writing URLs. Often times people use mod_rewrite through a .htaccess file. The problem with that is say you have a file 5 folders deep: /home/folder1/folder2/folder3/folder4/file-here Every time file-here is called, the server is checking all five folders for the existence of a .htaccess file. If it finds one, it has to open it and read it. The solution is to edit your configuration file (eg apache’s conf file) and put the mod_rewrite there. This way you save five file checks (and potentially one or two file reads) for every single hit.

This is of course just the tip of the iceberg - a lot of other things you can do to wring out more out of your server(s).

City-data.com. Look at its Compete.com numbers - estimated 3 million US visitors last month! Pretty close to Yelp.com, and well above of InsiderPages.com.

The site in itself is quite simplistic. Take (what appears to be public) city data, condense it nicely, and put it up. We had done something similar roughly 3 years ago (and City-data.com was around then too), but we never had as much data as they did. The site was getting roughly 10-15k uniques a day when we sold it off.

Interestingly, City-data.com was also one of the few sites (back in the day) that had Wikipedia content and ranked for it.

The owner himself uses this same approach to great success elsewhere. His Web Hosting Ratings site used to be #1 for ‘web hosting’ on Google - but he seemed to have stopped updating it and it has slid a bit. He also owns Faqs.org, a site I believe he purchased. That site has 209 DMOZ listings and 57 Yahoo listings (it used to have well in the thousands of DMOZ listings - getting a listing for every single RFC it published). These other sites of his were used to spread authority and google juice to his city-data.com website … which then took off and gained a lot of links.

City-data.com now has an astounding 1484 DMOZ listings, 1 Yahoo listing, and an insane 48,945 pages referencing that site on Wikipedia.

And don’t forget the forums - Threads: 203,075, Posts: 2,539,326, Members: 248,761

So there you have it. Another local-oriented site getting a lot of traffic, a lot of pull, a lot of stickiness (via its forum) … and being completely ignored by everyone in the local space.

So I got back today - I made a little bit of a blunder and we left the airport … when the flight was leaving in an hour. So we rushed through, cut the lines everywhere, and had time to grab a coffee before we got onto the plane.

I get home and Jacob from Bloggy Network is excited. Turns out Yahoo did some feature on ‘top us jobs’ (on the front page) and linked to search results for that phrase. What came up at #2? None other than one of our sites - Life Spy.

So the site ended up making an extra $1500 (give or take a bit) that it would have otherwise.

What really stunned me was the ‘quality’ of the traffic. The front page feature/search link sent us an extra ~45,000 pageviews. But the CTR on that page was around 15% - giving us a yummy CPM of roughly $35.00! To compare, the normal CTR/CPM are less than 1/4th for each.

So there you have it. One link made us over $1500 in just 12 hours. Looking forward to Yahoo doing it again (and I can only imagine what #1 would have done).

So I had two packages to be delivered today, and one tomorrow - all three by UPS.

The one by ground got here just fine. It was ETAed to arrive on Monday, but got Thursday.

The one that was sent by one day shipping - the idiot driver claimed my address didn’t exist. Of course when I called them up it turned out that the full address was there. In the last 10 times UPS has had to deliver something to me, twice they have claimed the address doesn’t exist, and one time they delivered the package to an apartment 2 blocks off.

I really don’t get what in the hell UPS is doing (and who they are hiring).

But today I got a first - a recycled tracking code!

So when I called to yell about my screwed up delivery, I had to yell about this too. It took like 40 minutes to resolve it … which consists of me having to drive to a place 40 minutes away to pick up both packages.

Sigh!

Update the Next Day: Evidently once they get into a groove of screwing you, they continue.

I was told I would get a call at around 10 am to confirm that the package (with the recycled tracking number) would be held for me. 10 am came and went whoosh. So I called, and the rep claimed they called me. Except of course I have caller ID - and there was no call. When I caught him with that, he claimed it said they called, but that no one of that name lived there, so they must have dialed it wrong. When I asked him why didn’t they bother to RE-DIAL then … he was left speechless.

And checking online it says ‘Out for Delivery’ … timed at 6:47 am.

If you ever hear about me sending something via UPS - be sure to kick me in the balls. Both result in the same level of physical and psychological pain.

ReachLocal - recipient of $67million in funding.

The footer includes a link to ReachLocal Directory … which is nothing more than an ODP directory.

What exactly is the point of having that?

UPDATE: Fixed the funding - I had erroneously noted it as $160 million. Someone from RL did contact me about this - I was told that it is just an old directory page … but I fail to see why it is linked from their front page still.

1. Things will tighten. While the USD seems to have stopped its freefall (3 months ago it was 1 USD = 0.98 CAD, now its upto 1.02), year over year it got brutally hurt. With high oil prices finally start to affect consumer behaviour, and the housing market meltdown creating a huge gap in people’s spending power, the US will see a legitimate correction/ouch/get your breath moment. I think this will cause some of the more no-revenue sites (even with sizable traffic) to feel some pain. This will finally cause …
2. The Great Offline to Online Local Migration. Local businesses are alrady starting to get unhappy with yellowpage publishers. There is enough evidence in front of everyone that less and less people are bothering with the yellow pages book anymore. Combined with the general tightening of things, enough small businesses will finally try to understand the internet and move their spending online. Which means …
3. The rise of the online local ad networks. Part of local’s difficulty (and accompanying beauty) is the granularity. Companies like ReachLocal, which are helping companies move money from offline to online are focusing on the major search engines. But there is a ton of local content online - from event sites to blogs to everything in between. I believe we will finally see the rise of online local ad networks that distribute ads across a swath of websites. A very sizable amount of traffic that is ignored by most are …
4. Domains (which will finally start to cool down). As things tighten, branding becomes more important. Which means generics like Food.com (great for traffic and SEO but harder to brand) won’t go skyrocketing like they were before. Or they may stay around the same … just like …
5. Google, Yahoo, MSN - nothing will change. Maybe AOL. But the Big Three will remain the Big Three (really people need to give Yahoo some credit for stealing Flickr when it did).

1. The previous record we had for # of pages crawled (for one site) by Google was roughly 430k. In the last 2 weeks we had one site that had over 500,000 pages hit by Googlebot. In 24 hours. Needless to say the server was a bit hot

2. I seem to be getting married in under a week. In fact the marriage is exactly on our 50 month anniversary. So after my 2 week Colombia excursion, expect more AWOLness.

3. This month we will break 2 million pageviews served in the local space. By the end of this year we should hit do 5 million a month.

So I’m back - spent 2 weeks in Colombia (the country). I went there for a wedding - my friend’s mother owns a beach-house there, and 15 of us crammed into that place. Not that I’m complaining - all the hotels there were $300+ per night, and the house was literally 10 seconds away from the beach.

The internet there was a bit of a chore - mostly because of damn forced localization. I can’t underline how difficult it was to go through a site like Hotmail or travel sites. At least Google had the sense to (after forwarding me to Google.com.co) to say (in plain english) ‘Go to Google.com’. The rest? Nothing. Even Sofitel.com forced me to the spanish version, refusing to let me access the english site.

Being smart and using IP targeting to redirect is a good idea. But you need to allow your visitors to access the other versions/areas if they so desire.

And another reason Google ‘wins’