Loren asks “will you switch to Ask?”

Loren Heiny asks the big question: will you switch to Ask?

I won’t.

Why not? Because of speed. Sorry, a KEY part of Google’s quality (and TechMeme’s, to tell the truth) is that Google’s results come back almost instantly. I just did a search for “Dave Winer” and Ask’s results took about two seconds to appear. Google’s appeared in less than half a second.

Now that I’ve seen the inside of part of a Google datacenter I understand just how tough it’ll be to beat. Google’s datacenters are just so amazing. It’s funny, since I was in that datacenter last Friday I’ve met a bunch of Googlers. None of whom have ever been in a Google datacenter.

Why is that funny? Because the datacenter is Google’s KEY DEFENSE against “new” competition.

From now on if anyone says they are gonna be better than Google at something I’m going to ask for a tour of their datacenter first.

If you don’t have datacenter excellence there’s NO WAY you’re going to beat Google at coming back fast.

It all starts in the datacenter. I’m not switching until someone demonstrates they can compete with Google in the datacenter. That said, it’s nice that Ask improved the UI and is bringing other kinds of searches onto its result page. I’ll keep trying it out, but I won’t switch to it.

87 thoughts on “Loren asks “will you switch to Ask?”

  1. “Why not? Because of speed”

    so if some search engine manages to bring in the *exact and most relevant* results you were looking 9/10 times but takes about 4 secs, you still wouldn’t switch?

    Like

  2. “Why not? Because of speed”

    so if some search engine manages to bring in the *exact and most relevant* results you were looking 9/10 times but takes about 4 secs, you still wouldn’t switch?

    Like

  3. It’s not the datacenter Robert, it’s the speed. One is not equal measure of the other.

    Google’s DC’s are not what won it breaking ground. The are what keep it in pole position (and of course a lot of UI intel).

    Like

  4. It’s not the datacenter Robert, it’s the speed. One is not equal measure of the other.

    Google’s DC’s are not what won it breaking ground. The are what keep it in pole position (and of course a lot of UI intel).

    Like

  5. To be fair, I’m not sure how you conclude that Ask.com doesn’t have equal or higher aptitude in data center operations. To the untrained eye (Robert) ANY data center looks like a marvelous thing. I’ve seen google’s space in a couple of different data centers. Excellent? Yes. Better than Microsoft, Ask, Amazon, eBay in data center operations? Not from what I could tell.

    Like

  6. To be fair, I’m not sure how you conclude that Ask.com doesn’t have equal or higher aptitude in data center operations. To the untrained eye (Robert) ANY data center looks like a marvelous thing. I’ve seen google’s space in a couple of different data centers. Excellent? Yes. Better than Microsoft, Ask, Amazon, eBay in data center operations? Not from what I could tell.

    Like

  7. Now, what if the Amazon web services were bolted onto an ubersearch algo? You’ve got scale, reliability, performance and high relevance without having laid out the jack for your own infrastructure…

    Mr. Andreesen talked on his blog recently about the advantages startups have available to them today that simply did not exist until recently. Open source and infrastructure have sure lowered the barriers to entry.

    Like

  8. Now, what if the Amazon web services were bolted onto an ubersearch algo? You’ve got scale, reliability, performance and high relevance without having laid out the jack for your own infrastructure…

    Mr. Andreesen talked on his blog recently about the advantages startups have available to them today that simply did not exist until recently. Open source and infrastructure have sure lowered the barriers to entry.

    Like

  9. seshadri: Google already hits the results I’m looking for at about 8 out of every 10 times. The real question is if you could improve it to, say, 10 out of 10, would I switch to a slower service? No. The extra 20% utility wouldn’t be worth the frustration of waiting for a page to load.

    Like

  10. seshadri: Google already hits the results I’m looking for at about 8 out of every 10 times. The real question is if you could improve it to, say, 10 out of 10, would I switch to a slower service? No. The extra 20% utility wouldn’t be worth the frustration of waiting for a page to load.

    Like

  11. Teknologist: >>Better than Microsoft, Ask, Amazon, eBay in data center operations? Not from what I could tell.

    I’ve seen the inside of some of Microsoft’s datacenters. They are pretty good, but not run as well as Google’s, sorry. Google’s are standardized and the racks are far better organized than what I’ve seen inside Microsoft’s.

    But, Microsoft’s pages do come up almost as fast as Google’s. Of course they have 1/6th as much traffic (or less) than Google does.

    Like

  12. Teknologist: >>Better than Microsoft, Ask, Amazon, eBay in data center operations? Not from what I could tell.

    I’ve seen the inside of some of Microsoft’s datacenters. They are pretty good, but not run as well as Google’s, sorry. Google’s are standardized and the racks are far better organized than what I’ve seen inside Microsoft’s.

    But, Microsoft’s pages do come up almost as fast as Google’s. Of course they have 1/6th as much traffic (or less) than Google does.

    Like

  13. “The extra 20% utility wouldn’t be worth the frustration of waiting for a page to load.”

    Currently when you get some results back for a query, do you scan the first 6-8 links or do you just click on the top one? If you do the former, then that would offset the delay …

    (I normally take a look at the first page of results to pick the most relevant one, and i thought that was pretty much the standard practice. )

    Like

  14. “The extra 20% utility wouldn’t be worth the frustration of waiting for a page to load.”

    Currently when you get some results back for a query, do you scan the first 6-8 links or do you just click on the top one? If you do the former, then that would offset the delay …

    (I normally take a look at the first page of results to pick the most relevant one, and i thought that was pretty much the standard practice. )

    Like

  15. It’s not about the DCs. It’s about scalability, structure, caching, optimization. It’s all about software.
    I bet my Windows runs faster on the same machine than yours. Optimization, cut the crap out, save queries.

    The response speed from Google was already reached in 2002 : watch this vid. You’ll see that traffic, number of queries and number of computers aren’t linear.
    Give any moron 450k servers. He’ll probably still be slower than Digg.

    Like

  16. It’s not about the DCs. It’s about scalability, structure, caching, optimization. It’s all about software.
    I bet my Windows runs faster on the same machine than yours. Optimization, cut the crap out, save queries.

    The response speed from Google was already reached in 2002 : watch this vid. You’ll see that traffic, number of queries and number of computers aren’t linear.
    Give any moron 450k servers. He’ll probably still be slower than Digg.

    Like

  17. No search engine touches google for speed, I will never use another engine, google reader and all the other features they have are great too.

    Like

  18. No search engine touches google for speed, I will never use another engine, google reader and all the other features they have are great too.

    Like

  19. Seshadri: problem is that I do the same searches for different reasons.

    For instance. Sometimes when I search for “Scoble” I’m looking for Maryam’s blog. Sometimes I’m looking for my Wikipedia page. Sometimes I’m looking for other things.

    There is no such thing as only one result that’s perfect every time. Chasing that will lead you down a rathole.

    Like

  20. Seshadri: problem is that I do the same searches for different reasons.

    For instance. Sometimes when I search for “Scoble” I’m looking for Maryam’s blog. Sometimes I’m looking for my Wikipedia page. Sometimes I’m looking for other things.

    There is no such thing as only one result that’s perfect every time. Chasing that will lead you down a rathole.

    Like

  21. Will I switch?

    Test START

    Well, when I search for my name, all the search engine in world brings back my site as first link as it has my name in it. Ask does not!

    Test END

    I will NOT!

    Like

  22. Will I switch?

    Test START

    Well, when I search for my name, all the search engine in world brings back my site as first link as it has my name in it. Ask does not!

    Test END

    I will NOT!

    Like

  23. Wow, I just searched Scoble on Ask and it actually took about two seconds, it was much quicker on google.

    Like

  24. Wow, I just searched Scoble on Ask and it actually took about two seconds, it was much quicker on google.

    Like

  25. Robert,

    I agree that the new Ask, cool as it is, is not yet compelling enough to switch from Google. Not only speed but I was mystified by the question of how they rank-ordered results. It seemed all but random.

    Wait a minute. Scratch the “all but.” It seemed random. The results weren’t chronological, didn’t seem to have any relevance ranking, and were a pretty mixed bag of accuracy.

    That said, I’ve come to a place where I use two search engines most days. Google is my primary engine but I’ll pop onto answers.com when I’m looking for depth on a particular topic. If I want to do a broad search, I use NoteTaker’s script that searches multiple engines, builds a notebook of the results including live pages, and removes dupes, etc. That’s a Mac-only product but it really is brilliant for this sort of thing.

    Like

  26. Robert,

    I agree that the new Ask, cool as it is, is not yet compelling enough to switch from Google. Not only speed but I was mystified by the question of how they rank-ordered results. It seemed all but random.

    Wait a minute. Scratch the “all but.” It seemed random. The results weren’t chronological, didn’t seem to have any relevance ranking, and were a pretty mixed bag of accuracy.

    That said, I’ve come to a place where I use two search engines most days. Google is my primary engine but I’ll pop onto answers.com when I’m looking for depth on a particular topic. If I want to do a broad search, I use NoteTaker’s script that searches multiple engines, builds a notebook of the results including live pages, and removes dupes, etc. That’s a Mac-only product but it really is brilliant for this sort of thing.

    Like

  27. Do they have donuts in the datacenter or something? You’re just so excited you used datacenter in each paragraph except the first — and I could tell you wanted to even then.

    C’mon, technology is the key, not the datacenter — or it’s fluffy muffins.

    So, let’s rewrite this the way is should be written.

    Replace “datacenter” with “snack bar” and “datacenters” with “donuts”. Now, there’s something to get excited about.

    Like

  28. Do they have donuts in the datacenter or something? You’re just so excited you used datacenter in each paragraph except the first — and I could tell you wanted to even then.

    C’mon, technology is the key, not the datacenter — or it’s fluffy muffins.

    So, let’s rewrite this the way is should be written.

    Replace “datacenter” with “snack bar” and “datacenters” with “donuts”. Now, there’s something to get excited about.

    Like

  29. No donuts in the datacenter. Heheh.

    Sorry, you can’t separate out the technology from the datacenter with these things. Put Google on Zooomr’s servers and it’d fall over just as quickly, no matter how well written the software is.

    Like

  30. No donuts in the datacenter. Heheh.

    Sorry, you can’t separate out the technology from the datacenter with these things. Put Google on Zooomr’s servers and it’d fall over just as quickly, no matter how well written the software is.

    Like

  31. Dan, is that the Dan Shafer I used to work with on CNET’s Builder.com Live? I miss talking with you. Gotta come down to Monterey for some sushi!

    Like

  32. Dan, is that the Dan Shafer I used to work with on CNET’s Builder.com Live? I miss talking with you. Gotta come down to Monterey for some sushi!

    Like

  33. Robert, I ran the same query=”dave winer” against both ask and google.

    For one, a user cannot conclude about the “speed” of the search engine by the “speed” of the resultset’s display rendering in a browser. Look closely and you would see the stylistic differences between the way Ask paints the screen (probably a decision made by their UI Design team) and Google displays it results.

    But yes, from the perspective of “speed” of “user experience”, google seems quicker.

    Here’s the thing – if you really want to compare their engine/infrastructure speeds – I suggest you compare the speed of “Google Suggest” and Ask’s Search Suggestions when you type any query.
    Try “Dave Winer” again!

    Like

  34. Robert, I ran the same query=”dave winer” against both ask and google.

    For one, a user cannot conclude about the “speed” of the search engine by the “speed” of the resultset’s display rendering in a browser. Look closely and you would see the stylistic differences between the way Ask paints the screen (probably a decision made by their UI Design team) and Google displays it results.

    But yes, from the perspective of “speed” of “user experience”, google seems quicker.

    Here’s the thing – if you really want to compare their engine/infrastructure speeds – I suggest you compare the speed of “Google Suggest” and Ask’s Search Suggestions when you type any query.
    Try “Dave Winer” again!

    Like

  35. Robert,

    When Google first started out it was probablly much smaller then Zoomr and has grown to the 450k servers or how many they have. The thing that Google has done is taken a Linux kernel and customized it to do nothing else then return search results fast. The tweaking of the kernel/drviers/code they use is what makes everything better. There is a reason they use Linux and the fact is they can tweak everything to optimize it. Remember I can make a Datacenter look awesome with mood lighting and everything if I wanted to, but that wouldn’t make it better then Google or MS or whatever. It’s the technology, not the look or feel of the datacenter.

    Another point to back this up: A company in my hometwon had an awesome looking datacenter, lots of blinking servers showing lots of activity, everything managed clean, cabling clean and ordered, everything was top shelf. The problem was the panels were just that. Server cases racked with blinking LEDs showing activity. The company was all a scam.

    Like

  36. Robert,

    When Google first started out it was probablly much smaller then Zoomr and has grown to the 450k servers or how many they have. The thing that Google has done is taken a Linux kernel and customized it to do nothing else then return search results fast. The tweaking of the kernel/drviers/code they use is what makes everything better. There is a reason they use Linux and the fact is they can tweak everything to optimize it. Remember I can make a Datacenter look awesome with mood lighting and everything if I wanted to, but that wouldn’t make it better then Google or MS or whatever. It’s the technology, not the look or feel of the datacenter.

    Another point to back this up: A company in my hometwon had an awesome looking datacenter, lots of blinking servers showing lots of activity, everything managed clean, cabling clean and ordered, everything was top shelf. The problem was the panels were just that. Server cases racked with blinking LEDs showing activity. The company was all a scam.

    Like

  37. “Put Google on Zooomr’s servers and it’d fall over just as quickly, no matter how well written the software is”

    You are over simplifying things and using incorrect inverse logic.

    Consider this analogy – Any ‘A’ list blogger needs to blog 4-5 times/day consistently. He would probably lose the A listing if he doesn’t. But you don’t consider the high frequency of posts as the difference between a ‘A’ lister and ‘Z’ lister. Do you?

    Like

  38. “Put Google on Zooomr’s servers and it’d fall over just as quickly, no matter how well written the software is”

    You are over simplifying things and using incorrect inverse logic.

    Consider this analogy – Any ‘A’ list blogger needs to blog 4-5 times/day consistently. He would probably lose the A listing if he doesn’t. But you don’t consider the high frequency of posts as the difference between a ‘A’ lister and ‘Z’ lister. Do you?

    Like

  39. Ridiculous post. Don’t post when you are drunk.

    Ask response time : for some queries, if the first three results are far more relevant than Google, then it’s worth it. What this means is that Ask can attract a niche of users.

    Google data centers : who cares. What’s so mysterious to see there?

    Google response time being the vibrant force against new competition : ridiculous. Google attracted audiences in late 90s for being better, not for being faster.

    Like

  40. Ridiculous post. Don’t post when you are drunk.

    Ask response time : for some queries, if the first three results are far more relevant than Google, then it’s worth it. What this means is that Ask can attract a niche of users.

    Google data centers : who cares. What’s so mysterious to see there?

    Google response time being the vibrant force against new competition : ridiculous. Google attracted audiences in late 90s for being better, not for being faster.

    Like

  41. @ #20 –

    What do you consider an “A-list” blogger? Technorati Top 2000?

    Then how do explain this blog (http://lagesse.org/index.php/2007/06/05/how-did-violent-acres-get-in-the-technorati-2000-in-under-8-months/) coming into the Top 1500 averaging less than one post per day?

    A LOT of the Top 2000 post once a day, on average. It is not about frequency – it is about quality.

    If it were based on frequency of posting I would be in the Top 100.

    Like

  42. @ #20 –

    What do you consider an “A-list” blogger? Technorati Top 2000?

    Then how do explain this blog (http://lagesse.org/index.php/2007/06/05/how-did-violent-acres-get-in-the-technorati-2000-in-under-8-months/) coming into the Top 1500 averaging less than one post per day?

    A LOT of the Top 2000 post once a day, on average. It is not about frequency – it is about quality.

    If it were based on frequency of posting I would be in the Top 100.

    Like

  43. I have just understood why you posted this crap. This is how you are asking Google to let you in and film the server racks.

    Sense of dignity be gone. Why don’t you IM them and keep it off the record instead of posting it?

    Like

  44. I have just understood why you posted this crap. This is how you are asking Google to let you in and film the server racks.

    Sense of dignity be gone. Why don’t you IM them and keep it off the record instead of posting it?

    Like

  45. Ask.com has a customizable skin functionality: user can select a background image for the search box.

    I was curious how large the image is because it will affect the amount of data travelling between user’s PC and Ask.com servers. Of course, the image is normally retrieved from the browser cache.

    Anyway, one of the images is around 250K, and the URL points to Akamai. So, at least for images, Ask.com is utilizing Akamai’s datacenters. But I’d guess they execute the actual search internally in the company’s servers. Heck, I don’t know: you can outsource almost anything nowadays.

    Like

  46. Ask.com has a customizable skin functionality: user can select a background image for the search box.

    I was curious how large the image is because it will affect the amount of data travelling between user’s PC and Ask.com servers. Of course, the image is normally retrieved from the browser cache.

    Anyway, one of the images is around 250K, and the URL points to Akamai. So, at least for images, Ask.com is utilizing Akamai’s datacenters. But I’d guess they execute the actual search internally in the company’s servers. Heck, I don’t know: you can outsource almost anything nowadays.

    Like

  47. > Because the datacenter is Google’s KEY DEFENSE
    > against “new” competition.

    A possible way to overcome this could be the use of
    peer-to-peer technology for search engines: http://www.faroo.com

    Like

  48. > Because the datacenter is Google’s KEY DEFENSE
    > against “new” competition.

    A possible way to overcome this could be the use of
    peer-to-peer technology for search engines: http://www.faroo.com

    Like

  49. I half think you were fishing for comments here, but nonetheless I want to comment, since I happen to live and breathe in the datacenter space. There’s a key distinction here between datacenter and infrastructure that I think everyone is tripping over. You’re saying a few things at once here in my opinion.

    1. Google’s datacenters are pretty. That hints at some great operational efficiencies and real attention to the details.

    2. Google’s infrastructure is great. They have selected systems that best suit their applications.

    3. Google’s search is fast, accurate, and reliable. It has been so for some time, so it would take more than an incremental difference (ie. a little bit faster, or a little bit more accurate) to move you.

    Does datacenter tidiness mean the application will run faster? Strictly speaking, no. However, the fact that they have neat and tidy DCs suggest that they have their act together around IT processes. When they need to scale up, out or respond to major infrastructure issues, they likely understand exactly what’s required and can perform with ease. Based on my understanding of google’s business model, that is essential for them. Scaling up and out efficiently and minimizing downtime is critical.

    Now, that being said did they show Robert the cabling closet on the developers floor? Probably not, I’m sure it was one of their showcase production floors.

    Like

  50. I half think you were fishing for comments here, but nonetheless I want to comment, since I happen to live and breathe in the datacenter space. There’s a key distinction here between datacenter and infrastructure that I think everyone is tripping over. You’re saying a few things at once here in my opinion.

    1. Google’s datacenters are pretty. That hints at some great operational efficiencies and real attention to the details.

    2. Google’s infrastructure is great. They have selected systems that best suit their applications.

    3. Google’s search is fast, accurate, and reliable. It has been so for some time, so it would take more than an incremental difference (ie. a little bit faster, or a little bit more accurate) to move you.

    Does datacenter tidiness mean the application will run faster? Strictly speaking, no. However, the fact that they have neat and tidy DCs suggest that they have their act together around IT processes. When they need to scale up, out or respond to major infrastructure issues, they likely understand exactly what’s required and can perform with ease. Based on my understanding of google’s business model, that is essential for them. Scaling up and out efficiently and minimizing downtime is critical.

    Now, that being said did they show Robert the cabling closet on the developers floor? Probably not, I’m sure it was one of their showcase production floors.

    Like

  51. If I search for Salubri on Ask – I get my blog up Top result. I’m very happy. 🙂 Woohoo… Salubri is a VERY commonly used “non word” on the internet (9000+ results on ask) as it is the name of a tribe of vampires in a game. I’m delighted that Ask puts me top of the list…
    Unfortunately I do not search for my own stuff all that regularly so I tried another…
    A blogspot.com blog by a friend of mine (more posts and more active till recently) – nope… First result is a (seemingly unrelated) advert and the second is a post on my blog that links him… Cool more traffic for me… Hmmm blogspot.com = blogger.com = ???… Curious!
    If I search for “lifesyncs” (a friend’s blog on blogger lifesyncs.blogspot.com) I get 2 results… A website by yours truly that refers to his blog and something altogether different… That’s wierd… Wait – theres a name on his blog “Synchronicity” hmm OK lets try “synchronicity movies spider-man disappointing” (that can’t be that common and he used all those phrases in one post on his blogrecently) Nope… 40 results but NOTHING related to his blog? How about “Synchronicity Sligo” he’s from Sligo and mentions it in posts… Nah – I give up at page one where there are LOADS of ads and not alot to do with synchronicity… Blogger.com not really a priority? I’m just saying… *shrug*
    Google returned the blogs I was looking for at top of the list for all those queries by the by… Except mine (I’m ok with that – the Salubri vampires had the name first and outnumber me 9000 sitees to 1) if I google “Salubri blog” I’m back to the top.

    Not that where my friends or I place in searches really matters a darn – just curious how the blogger blogs seems to be suffering on Ask…

    Oh – and the interface isn’t as nice as Technorati…

    Like

  52. If I search for Salubri on Ask – I get my blog up Top result. I’m very happy. 🙂 Woohoo… Salubri is a VERY commonly used “non word” on the internet (9000+ results on ask) as it is the name of a tribe of vampires in a game. I’m delighted that Ask puts me top of the list…
    Unfortunately I do not search for my own stuff all that regularly so I tried another…
    A blogspot.com blog by a friend of mine (more posts and more active till recently) – nope… First result is a (seemingly unrelated) advert and the second is a post on my blog that links him… Cool more traffic for me… Hmmm blogspot.com = blogger.com = ???… Curious!
    If I search for “lifesyncs” (a friend’s blog on blogger lifesyncs.blogspot.com) I get 2 results… A website by yours truly that refers to his blog and something altogether different… That’s wierd… Wait – theres a name on his blog “Synchronicity” hmm OK lets try “synchronicity movies spider-man disappointing” (that can’t be that common and he used all those phrases in one post on his blogrecently) Nope… 40 results but NOTHING related to his blog? How about “Synchronicity Sligo” he’s from Sligo and mentions it in posts… Nah – I give up at page one where there are LOADS of ads and not alot to do with synchronicity… Blogger.com not really a priority? I’m just saying… *shrug*
    Google returned the blogs I was looking for at top of the list for all those queries by the by… Except mine (I’m ok with that – the Salubri vampires had the name first and outnumber me 9000 sitees to 1) if I google “Salubri blog” I’m back to the top.

    Not that where my friends or I place in searches really matters a darn – just curious how the blogger blogs seems to be suffering on Ask…

    Oh – and the interface isn’t as nice as Technorati…

    Like

  53. Oh – the first blogspot.com blog with the odd name that only has 2 matches on Ask is sinisilma.blogspot.com – just so you can test it out on Ask and see…

    Like

  54. Oh – the first blogspot.com blog with the odd name that only has 2 matches on Ask is sinisilma.blogspot.com – just so you can test it out on Ask and see…

    Like

  55. @7 “but not run as well as Google’s, sorry.”

    You know that…how? Have you reviewed that ops procedures of both? If so, what leads you to your conclusion? You see one small part of Google’s datacenter and one part of Microsoft’s and come to this conclusion? GREAT analysis!

    Like

  56. @7 “but not run as well as Google’s, sorry.”

    You know that…how? Have you reviewed that ops procedures of both? If so, what leads you to your conclusion? You see one small part of Google’s datacenter and one part of Microsoft’s and come to this conclusion? GREAT analysis!

    Like

  57. “A LOT of the Top 2000 post once a day, on average. It is not about frequency – it is about quality”

    Exactly. just like high frequency is *one* of the traits, DCs are one such thing for a web 2.0 company.

    Like

  58. “A LOT of the Top 2000 post once a day, on average. It is not about frequency – it is about quality”

    Exactly. just like high frequency is *one* of the traits, DCs are one such thing for a web 2.0 company.

    Like

  59. “but not run as well as Google’s, sorry.”

    Yet another baseless Scoble proclamation. Scoble, who knows absolutely nothing about data centers, feels qualified to make pronouncements regarding them. Scoble, you don’t know yourself very well; from reading this blog, it’s clear to me that you’re very easily “wow’ed” (as evidence, all of your “(Insert gadget or web service here) Rocks!!” posts for mediocre products.

    Just because you (especially you) were wow’ed by a bunch of servers in a room doesn’t mean anything at all.

    Like

  60. “but not run as well as Google’s, sorry.”

    Yet another baseless Scoble proclamation. Scoble, who knows absolutely nothing about data centers, feels qualified to make pronouncements regarding them. Scoble, you don’t know yourself very well; from reading this blog, it’s clear to me that you’re very easily “wow’ed” (as evidence, all of your “(Insert gadget or web service here) Rocks!!” posts for mediocre products.

    Just because you (especially you) were wow’ed by a bunch of servers in a room doesn’t mean anything at all.

    Like

  61. Brevin: you’re right. +I+ don’t know much about running data centers (although I did the wiring on Winnov’s datacenter back in the mid-1990s). But I’ve spent quite a few hours with people who DO run datacenters.
    Why did I get so excited about Google? Have you ever been in a professionally-run datacenter? I guess not if you don’t get that their equipment is easier to manage, lower cost to acquire, more power efficient, and architected (and wired) better. Who said that? Me? No. Raju at Zoho who is in awe of Google’s datacenter execution too.

    Like

  62. Brevin: you’re right. +I+ don’t know much about running data centers (although I did the wiring on Winnov’s datacenter back in the mid-1990s). But I’ve spent quite a few hours with people who DO run datacenters.
    Why did I get so excited about Google? Have you ever been in a professionally-run datacenter? I guess not if you don’t get that their equipment is easier to manage, lower cost to acquire, more power efficient, and architected (and wired) better. Who said that? Me? No. Raju at Zoho who is in awe of Google’s datacenter execution too.

    Like

  63. I agree 100% with Robert–speed is VERY important. WE have more servers than we need and that’s why we are soooooo fast right now. Our pages are much longer and larger than Google or Ask’s, but we are right behind Google and ahead of Ask in terms of speed (or close to ask).

    I suggest going to the home page and trying a bunch of links. We’re having big debates now if we should add images to pages or not due to speed.

    Very hard decisions… some folks are obsessed with speed like Robert, some folks want more content/images/video. Some folks want different things are different times.

    ugh… I’m glad we are in Alpha and have a couple of years to figure this out!

    Jason
    (CEO of Mahalo.com)

    Like

  64. I agree 100% with Robert–speed is VERY important. WE have more servers than we need and that’s why we are soooooo fast right now. Our pages are much longer and larger than Google or Ask’s, but we are right behind Google and ahead of Ask in terms of speed (or close to ask).

    I suggest going to the home page and trying a bunch of links. We’re having big debates now if we should add images to pages or not due to speed.

    Very hard decisions… some folks are obsessed with speed like Robert, some folks want more content/images/video. Some folks want different things are different times.

    ugh… I’m glad we are in Alpha and have a couple of years to figure this out!

    Jason
    (CEO of Mahalo.com)

    Like

  65. I know each time I moved my comparatively paltry 120 servers from one data center provider to another (forced each time, due to bankruptsy or DC shutdown) we GREATLY improved the “visual” tidiness, but the shit still ran as fast/slow as it did previously. But it was dust free (yes, there IS dust even in a climate controlled data center) for a while, and the wiring was prepped slowly and deliberately. It impressed the ceo (scoble-types), but did nothing to affect performance or reliability. Then over time, as emergencies happen and you rush servers into service… the perfect visual becomes muddy once again.

    Like

  66. I know each time I moved my comparatively paltry 120 servers from one data center provider to another (forced each time, due to bankruptsy or DC shutdown) we GREATLY improved the “visual” tidiness, but the shit still ran as fast/slow as it did previously. But it was dust free (yes, there IS dust even in a climate controlled data center) for a while, and the wiring was prepped slowly and deliberately. It impressed the ceo (scoble-types), but did nothing to affect performance or reliability. Then over time, as emergencies happen and you rush servers into service… the perfect visual becomes muddy once again.

    Like

  67. One more thing… I havent seen (knowingly) a microsoft installation in a data center. However, I would have to assume they have an INCREDIBLE advantage in waiting until NOW to really start their data center buildouts. THey aren’t moving used servers and disk arrays into those new digs.

    Like

  68. One more thing… I havent seen (knowingly) a microsoft installation in a data center. However, I would have to assume they have an INCREDIBLE advantage in waiting until NOW to really start their data center buildouts. THey aren’t moving used servers and disk arrays into those new digs.

    Like

  69. Teknologist: part of Microsoft was — until a few weeks ago — housed in the same datacenter as Zoho. They moved out. You telling me they didn’t move those servers into their new data center? Uh, huh. You forget that Hotmail has 200 million people on it. You don’t just “throw out” that kind of data center and start again. And that’s only a very small piece of Microsoft’s datacenter.
    Oh, and did your datacenter have thousands of computers in it?
    Did all of your computers slide out for easy replacement? Did all your computers have the same processor, hard drive, power supply, fan?
    Were your computers open to the air without a case around them so they cooled more efficiently and required less fan power?
    Were they as cheap to build as Google’s?
    If so, I’d love to have your datacenter on my show and meet the guy who built it.
    The guy who built Zoho’s (more than a million of equipment) is in awe of Google’s data center execution. So are other people I know who know about datacenters.

    Like

  70. Teknologist: part of Microsoft was — until a few weeks ago — housed in the same datacenter as Zoho. They moved out. You telling me they didn’t move those servers into their new data center? Uh, huh. You forget that Hotmail has 200 million people on it. You don’t just “throw out” that kind of data center and start again. And that’s only a very small piece of Microsoft’s datacenter.
    Oh, and did your datacenter have thousands of computers in it?
    Did all of your computers slide out for easy replacement? Did all your computers have the same processor, hard drive, power supply, fan?
    Were your computers open to the air without a case around them so they cooled more efficiently and required less fan power?
    Were they as cheap to build as Google’s?
    If so, I’d love to have your datacenter on my show and meet the guy who built it.
    The guy who built Zoho’s (more than a million of equipment) is in awe of Google’s data center execution. So are other people I know who know about datacenters.

    Like

  71. Robert, if Microsoft were simply moving existing servers to to those brand-new data centers… that would be tantamount to misleading their investors. They simply aren’t filling those DCs up with used equipment. SURE, there is some legacy stuff. But the vast majority is new and are there for future growth. I didn’t mean to imply that not a single used machine would be moved. Sorry to confuse.

    > I’d love to have your datacenter on my show

    My work, as I noted, involved around 120 servers. And we didn’t have the luxury of capital to hit 120 on day 1. More like 6 years to get there. Still, even within that incrementally build server farm… we used a grand total of FOUR server hardware models, all from the same manufacturer. There were two box sizes (large for databases, and small for ‘everything else’). And its not a BAD thing that our manufacturer incremented the model numbers when they did upgrades. Don’t tell me that google has the same hard drives, etc. I call BS. They’ve had their site running for FAR too many years, for that to be the case.

    Like

  72. Robert, if Microsoft were simply moving existing servers to to those brand-new data centers… that would be tantamount to misleading their investors. They simply aren’t filling those DCs up with used equipment. SURE, there is some legacy stuff. But the vast majority is new and are there for future growth. I didn’t mean to imply that not a single used machine would be moved. Sorry to confuse.

    > I’d love to have your datacenter on my show

    My work, as I noted, involved around 120 servers. And we didn’t have the luxury of capital to hit 120 on day 1. More like 6 years to get there. Still, even within that incrementally build server farm… we used a grand total of FOUR server hardware models, all from the same manufacturer. There were two box sizes (large for databases, and small for ‘everything else’). And its not a BAD thing that our manufacturer incremented the model numbers when they did upgrades. Don’t tell me that google has the same hard drives, etc. I call BS. They’ve had their site running for FAR too many years, for that to be the case.

    Like

Comments are closed.