Philip Greenspun’s Weblog

Folks:

I’d like to spec out the required front-end server hardware for a database application with the following characteristics:

• Oracle server in the back end with at least 30 CPUs (existing machine; not sure how many cores in each CPU) and a lot of RAM
• Microsoft IIS/VB.NET Active Server Pages for the front-end computers (with load balancer)
• typical user session includes roughly the following workflow (let’s say 6 pages total)
  1. authenticate
  2. query Oracle to get enough information to build a form
  3. upon receiving the submitted form, send as many as 30 insert statements to Oracle

During a peak period we could conceivably get as many as 45,000 users logging in during a 30 minute period so that’s 270,000 pages to be served within 30 minutes or roughly 150 pages per second (corresponding roughly, perhaps, to the peak load on a server handling 4.5 million pages per day with a slightly uneven demand).

Let’s assume that Oracle is sufficiently well tuned and supported by hardware that it responds to all of these queries and transactions very quickly and does not become a bottleneck. (I recognize that this is a big assumption, but I would like to treat any Oracle issues in a separate posting.)

Here are my questions:

1. what is the best way to configure IIS/.NET to maximize database performance? The database won’t be doing much else during these peak periods and my assumption is that there is no advantage to having more simultaneously connected Oracle users than the number of CPU cores on the Oracle server. Does that make sense?
2. what kind of load can a $2000 Windows server support? The real work in this application is being done by Oracle on the monster Unix box in the middle. So all that the Windows server would have to do is interpret the Active Server Pages, manage a pool of database connections, and send complete pages back to users via HTTPS (note the SSL component). How many SSL pages per second would you expect the machine to be able to deliver? [For concreteness, I priced up a Dell R510 with dual Intel Xeon E5630 (2.5 GHz; 8 cores total from the two chips) and 16 GB of RAM. It came out to $2000 (admittedly with no disk or OS, but I guess you can’t have everything for $2000!). Anyway, let’s assume 8 cores and 16 GB of RAM.]
3. Is it easy to configure the latest version of IIS/.NET to open no more than, say, 8 connections to the DBMS?

Thanks in advance to the Windows/IIS/ASP.NET heroes who answer!

[Note: This was edited down from a posting that had some distracting stuff about tuning Java, which resulted in a very unfocused discussion. I have had the moderator remove the non-IIS/ASP comments and will save them for a later posting regarding Oracle/Java.]

Full post, including comments

Folks:

I’ve drafted a book review of Academically Adrift and would appreciate comments/corrections.

Thanks.

Full post, including comments

I just finished The Book of CSS3: A Developer’s Guide to the Future of Web Design and it gives a good survey of the capabilities of CSS3, e.g., a 1980s TV station-style flying logo in three dimensions (and, more usefully, capabilities for telling the browser to lay text out in multiple columns; see http://thebookofcss3.com/resources/chapter-7/ for examples (don’t seem to work that well in Chrome; the columns don’t resize when I adjust the browser window size)). It would also be a good reference. But the examples in the book seem contrived and overall I think it would be a lot easier to learn CSS3 with a book arranged project-by-project. Has anyone seen a good case study-oriented book on CSS? Or perhaps a Web site.

Full post, including comments

The Rational Optimist: How Prosperity Evolves gives an inspiring list of predicted catastrophes that have failed to occur. I’m going to limit the list to ones that have been predicted during my lifetime (born 1963).

Cancer. Rachel Carson predicted a big increase in cancer due to due to DDT, which would cause “practically 100 per cent of the human population to be wiped out from a cancer epidemic in one generation.” Paul Ehrlich, a MacArthur genius, also predicted doom from DDT: “The U.S. life expectancy will drop to forty-two years by 1980, due to cancer epidemics.”

(another) Nuclear War. It once seem inevitable that we would have a species-ending nuclear war with the Soviet Union.

Famine. Environmentalist Lester Brown predicted imminent famine in 1974, 1981, 1984, 1989, 1994, and 2007, in a 1967 book titled Famine, 1975!, and by MacArthur genius Paul Ehrlich in a 1968 book The Population Bomb (repeated, but without a predicted date, in a 2008 book, The Dominant Animal).

Exhaustion of minerals, oil, etc. We did indeed use up all of the known reserves of zinc, gold, tin, copper, oil and natural gas by 1992, just as predicted in the Club of Rome’s 1970s bestseller Limits to Growth. But then we found some more.

Air Pollution. Flying a small plane into the Lower 48 from Alaska or the Caribbean, I’m often amazed at how brown the air looks hanging over Washington State or Florida. On the other hand, it was supposed to be much worse. “In 1970, Life magazine promised its readers that scientists had ‘solid, experimental and theoretical evidence’ that ‘within a decade, urban dwellers will have to wear gas masks to survive air polluion… by 1985 air pollution will have reduced the amount of sunlight reaching earth by one half.'”

Plague (AIDS). “The number of deaths from [AIDS] has been falling since 2005.”

Plague (Flu). Ridley claims that the 1918 flu epidemic was unique due to the environment of soldiers in trenches. A successful modern flu needs its victims to be well enough to walk around, go to work, and spread the disease further.

Global cooling. If not for all of the dinosaur blood that we’ve pumped out of the ground and set on fire, we would in fact be going (slowly) into another ice age. This frightened journalists and their readers in the 1970s and a lot of ink was spent on dire warnings about climate change similar to today’s articles, except that the promised temperature trend was opposite.

Ridley suggests that we count our blessings. The last few years have seemed to offer a lot of lessons about human hubris. Here are some of the things that we’ve learned we can’t do safely: (a) drill for oil in mile-deep water (Deepwater Horizon), (b) build nuclear power plants on tsunami-prone coastlines (Fukushima), (c) build the world’s biggest airliner and engines (Qantas Flight 32), (d) protect cities that are below sea level from flooding (Katrina in New Orleans). On the other hand, none of this seems to stop the human population from expanding and, overall, from enjoying a better standard of living than previous generations.

[June 8 update: NYT carries a Thomas Friedman column “The Earth is Full” saying more or less the exact opposite of what Ridley wrote. They both have great credentials (Ridley has a doctorate in science; Friedman was smart enough to marry a billionaire heiress) so it is tough to know whom to believe.]

Full post, including comments

Steve Jobs unveiled the iCloud file syncing system only a few hours ago, yet already there are imitators who have managed to run with Apple’s ideas. The two copycats that come most readily to mind are Dropbox.com and Google Docs (like iCloud, a cloud-based service that stores files, but the Google programmers, apparently within just a few hours, also managed to develop a browser-based editing interface to some of those files).

Has anyone figured out what the original innovative Apple service does that these imitators haven’t yet managed to copy? I.e., what can I do with iCloud that I can’t do with Dropbox.com?

Full post, including comments

In reading opinions regarding the U.S. and state economies from professional economists and in talking to these folks face-to-face I’ve never heard any of them say anything clear enough for a voter to act upon. For example, suppose that the U.S. owed 800 percent of GDP, like Greece (see this NY Times article (the U.S. figure is around 500 percent, though we’re not as centralized as European countries and there is almost surely a lot of state-to-state variation)). An accountant would say “You’re spending more than you earn. You have to stop or you will run out of money.” After hearing that, voters and politicians might be able to get together and agree on some spending cuts (not the trivial ones that they’ve managed so far). Perhaps even Californians would be able to agree that their cities could be run into bankruptcy just as competently by managers who earned no more than the President of the United States earns, that a qualified fire chief could be hired for only the same salary as the U.S. Secretary of Defense, and that a police lieutenant need not receive 3X as much as a U.S. Army infantry lieutenant in combat in Afghanistan.

Instead of accountants in public discourse, however, we have economists. So our problems are not recognized as arithmetic challenges, e.g., “How can a society with a median wage of about $16 per hour afford to pay local and state government workers $100-200 per hour (and then pay them for another 50 years after they retire)?” or “How can teenagers who score lower than their Chinese counterparts on every measure of educational attainment be relied upon to pay for 130 million poor and older Americans to receive unlimited medical services in the world’s most expensive health care system?” (see this chart for how Medicare will have 80 million beneficiaries in 2030 (I added in another 50 million for Medicaid (chart)).

The problems are instead categorized as amenable to complex solutions that only someone with a Ph.D. in economics can understand. Or perhaps they are cast as political problems, e.g., “If not for the presence of a handful of heretics amongst voters and in the legislature, we could pass new laws that would magically double our wealth” (I have some extra faith in Massachusetts because one party has controlled our legislature for decades and when we run out of cash there is less chance that they will waste time trying to pin the blame on the handful of Republicans who show up to the State House every now and then to collect a paycheck (but do nothing else)). But oftentimes economists are politically oriented and contribute to the relabeling of “not enough cash” to “too many people in Party X”.

So instead of the sober accountant showing up on TV or in a news article saying “You aren’t rich enough to do stuff like this” or “All of your wealth was siphoned off by the following cronies of the current rulers” we have Economist A saying “Really the U.S. is in great shape if only we printed more money here, had one government agency issue some bonds there, had another government agency buy those bonds, and changed some assumptions in making projections about the costs of Programs X, Y, and Z”. That would be confusing enough except that Economist B then comes on and says “No, the government shouldn’t be printing money and we need these other assumptions in our projections.” The contradictory and complex opinions result in paralysis and inaction where the accountant’s simple warning would have been heeded. Perhaps worse, people hear what they want to hear, in which case a complex macroeconomic argument will register as “We can all be wealthier even if we’re studying less in school and then working fewer hours and fewer years.”

So could we increase our society’s wealth if we sent all of our economists off on a three-year holiday?

Full post, including comments

If you’re an expert Web developer with Python experience and want to move to the Imperial City, this database-backed Web developer opening in Washington, D.C. should be appealing. You’ll get paid reasonably good money and have a chance to work with Nathan Cobb, the physician who founded QuitNet, which has helped literally hundreds of thousands of people quit smoking. QuitNet is a good reference for when folks ask “Has anyone ever done anything useful with social media?”

Cobb is notable for learning about Facebook and Web 2.0 in 2004 then developing a time machine in which to travel back to 1995 and start his online community with many Facebook-like features.

The job should be secure since the money is ultimately coming from the $206 billion Tobacco Master Settlement Agreement between the cigarette companies and the states. [Most of those states will probably soon need to start sending every retired public employee a free carton of cigarettes every week in hopes of reducing their pension liabilities.] Not that you’ll really need the cash since all of your entertainment will be free (museums, lectures, concerts, recreation, etc., all paid for by collecting taxes from people in Ohio, Michigan, and other struggling states). You’ll be getting good experience building applications that plug into all of the popular social networking sites. You’ll be living in a city where there are literally 10 jobs for everyone with a modicum of skill, so if you don’t like it you will have your choice of places to work next.

[In case you’re a single male Python nerd tempted to move to Silicon Valley, keep in mind that there is a huge surplus of single men in the San Francisco Bay area whereas the opposite situation applies in Washington, D.C. (see Boston Globe (written by a journalist and therefore not adjusted for overall population size). Would you rather be by yourself fighting California traffic or with a lovely young female government employee riding the Metro? Remember that your government worker companion will be well-paid and she won’t have to work long hours so she’ll have plenty of time for recreation.]

Full post, including comments

I’m halfway through The Rational Optimist: How Prosperity Evolves by Matt Ridley, an English science journalist. The book covers 200,000 years of human history, but this blog posting concerns just one chapter “The feeding of the nine billion”.

Have you chuckled at the apparent inconsistency of a neighbor who drives a 7,000 lb. pavement-melting SUV to Whole Foods and then buys organic produce? It turns out that there is no inconsistency. She is destroying the planet with her SUV and with her purchases of hard-to-grow organic food.

Ridley notes that with genetically engineered crops, synthetic fertilizers, and Roundup to control weeds, the trend of feeding ever more people with less land could be continued. The biggest obstacle to returning land to its wild state is organic farming. Currently we are using 38 percent of the Earth’s land for growing food or grazing animals; at 1961 levels of productivity we would need to be using 82 percent of the land.

Organic farmers won’t use genetically engineered crops, so they spend a lot more time and energy fighting pests. Organic farmers won’t use Roundup and other herbicides, so they plow the weeds under, which kills a lot of small animals and loosens the soil enough that it erodes (or sometimes they resort to flame-throwers). Organic farmers won’t use standard fertilizer, but only manure from cows, which means we’ll need a lot more cows running around.

Organic cotton is an especially hard-on-the-Earth product, according to Ridley. Standard industrial cotton has Bacillus thuringiensis (“bt”) genes mixed in and these kill pests, cutting the need for sprayed pesticides in half.

Who knew that “sustainable” would mean a polyester shirt and a bag of Fritos?

[Update: Just a few days after this posting, a German E. coli outbreak that killed 22 people has been blamed on organic bean sprouts (story).]

Full post, including comments

The continuing decline of U.S. house prices has been in the news lately, along with articles about how people love to rent. In some parts of the country, buying is definitely cheaper than renting, especially when you consider the 4-5 percent interest rates being offered. Can it be rational under those circumstances to rent?

One factor that has been covered is the high transaction cost of selling a house, perhaps 10 percent of the house price (5-6 percent real estate commission plus the cost of leaving the place vacant for 3-12 months). That’s equivalent to 1-3 years of rent. What if the person knows for sure that he or she will stay at least ten years and therefore it genuinely might be cheaper to buy. Why doesn’t he or she get out a checkbook?

Let’s consider an adjustment for the risk that, when it comes time to sell the house, there won’t be any jobs in the city or state and therefore the value of the house will be zero. If you’re a working-age renter, by definition your apartment is near a job and within a reasonably functional economy. Otherwise you wouldn’t be there paying rent! Between 1970 and 2008, a home buyer probably would not think to discount a house for the possibility that an entire city or state economy would essentially fall apart. Such things had not happened in recent memory (though they had happened, e.g., in Lowell and Lawrence, Massachusetts when the textile mills shut down and moved south; population fell and houses became surplus).

The potential home buyer today has seen pictures of Detroit, with former neighborhoods being gradually reclaimed by Nature or plowed under into farmland. Recognizing that his or her own city could become like that in 20 years time, the buyer will factor that into the price he or she is willing to pay. In the event of a Detroit-style decline, the house becomes worthless and the cost of ownership for 10 years or so effectively tripled (10 years x 5 percent is approximately equal to 50 percent of the home’s value, then add another 100 percent for the cost of throwing the house away). Suppose the buyer thinks that this has a 20 percent probability of happening. Given a typical person’s risk aversion, that might reduce the market-clearing price for a house by 25 percent.

I’m not quite sure how to test this theory. There are some parts of the U.S. where the risk of economic decline is much lower than others. For example, it is hard to imagine how Washington, D.C. could fail to prosper, even if much of the rest of the nation is impoverished. So the “discount for risk of write-off” should be near zero in Northwest DC and Bethesda, Maryland (though some neighborhoods of D.C. did decline to near worthlessness within recent memory). Manhattan seems also like a place where it is very likely there will be jobs. Perhaps Santa Monica and the nicer parts of San Francisco/Silicon Valley too (the data are pretty coarse, though, and the house market for Los Angeles overall may not correlate that well with Santa Monica when times are tough). Looking at the buy/rent ratio we would expect it to be higher in places with less of economic collapse.

http://economix.blogs.nytimes.com/2011/05/10/rent-vs-buy-a-longer-list/

has some suggestive data. Cities that seem at risk of being abandoned altogether, such as Detroit and Cleveland, are indeed theoretically very cheap places to buy. Washington, D.C. is expensive as well as some geographically blessed places such as San Francisco, Seattle, Orange County, and Honolulu. New York City, however, is only at an average ratio (maybe because Manhattan is not broken out separately?).

Full post, including comments

Friends have been asking me to comment on the latest news regarding the June 1, 2009 crash of Air France 447, e.g., this CNN article. I dug up my June 9, 2009 posting on the subject (pure speculation at the time) and, as the black box story unfolds, would appreciate readers grading my success.

Full post, including comments