Another hazard of Laptopistan - slow WiFi

This is hardly unexpected, but one of the side effects of everyone using Starbucks as a de facto workplace and internet cafe is that their WiFi is heavily stressed. People watching online video, using Remote Desktop (like me), probably even BitTorrent. I don't know what sort of routers they put in these coffee shops, but they often seem to be unequal to the load they're handling.

This morning I'm in a Starbucks, and my connection went from perfectly fine, at 8:30 AM before anyone was here, to barely usable now that I'm surrounded by laptops. One more reason for actual workspace.

Sliding Window partitioning

Implemented my first "sliding window" partitioned table. Wound up being easier than I expected, especially since the table will only have 2 partitions for now - Current and Archive.

Most sliding window schemes I've seen out there on the intertubes make use of a date as the partition key and update the sliding window based on the current date. In this case, the data comes in somewhat sporadically, so although it's date-based (a month key), I made the partitioning data-driven instead of date-driven. So for starters we have a RANGE LEFT scheme like this:

Partition Month Key

Current -

Archive 88

Say we're currently on month 90. So months 89 and 90 are in Current, and 1-88 are in Archive. When the sliding window proc runs, it does essentially this:

SELECT @NewBoundary = MAX(monthKey)-2 FROM tbPartitionedFoo

SELECT @CurBoundary = CAST(prv.value as smallint) FROM sys.partition_functions AS pf

JOIN sys.partition_range_values as prv

ON prv.function_id = pf.function_id

WHERE pf.name = 'partfn_foo'

(If there was more than one partition range value, a MIN() or something would be needed to grab the appropriate value.)

IF @NewBoundary > @CurBoundary

BEGIN

ALTER PARTITION FUNCTION partfn_foo() SPLIT RANGE (@NewBoundary)

ALTER PARTITION FUNCTION partfn_foo() MERGE RANGE (@CurBoundary)

END

So when we start getting monthKey 91 data and the maintenance proc runs, it will do this:

IF 89 > 88

BEGIN

ALTER PARTITION FUNCTION partfn_foo() SPLIT RANGE (89)

Now we have 3 partitions:

Partition Month Key

Current -

temp 89

Archive 88

ALTER PARTITION FUNCTION partfn_foo() MERGE RANGE (88)

END

And now we're back to 2:

Partition Month Key

Current -

Archive 89

Python Gets Things Done

A quick example of why I've grown to love Python.

My girlfriend had a laboriously scanned document in two pieces that needed to be combined. I had a .NET command line utility that I wrote a couple years ago that does just that, but because one of the scanned pages was a different size, it got cut off in the resulting output file. So I could either try to dig up some docs on the library I used, play around with Intellisense in Visual Studio, or see if there was a Python library that might work.

2 minutes of Googling later I had pyPDF, and maybe 10 minutes after that, I had reset the mediaBox on the outsized page, written it to a new pdf file, and emailed it off. Q.E.D.

DATEs aren't INTs, at least not any more

I updated a database column today to utilize the SQL 2008 DATE type, upgrading from a SMALLDATETIME which always had a 00:00:00 hh:mm:ss component. I was fairly certain that this would not break any existing code, so I did not perform an exhaustive code search. My mistake.

A function broke with the following message:

Msg 206, Level 16, State 2, Line 1

Operand type clash: date is incompatible with int

"Strange," I thought. "How would an int be getting used as a date?"

Of course there's the basic method of datetime construction from a string: CAST("2012-01-01" as datetime), but that doesn't involve any INTs. Did someone try to use a similar INT (20120101) in place of the string? No, because that would cause an arithmetic overflow...

Well, it turns out that there's another way of using INTs as DATETIMEs:

select cast(CAST('2012-01-10' as datetime) as int)

Apparently this doesn't work any more:

select cast(CAST('2012-01-10' as date) as int)

And this is what was being performed in the function in question:

...

CASE WHEN io_end > 40947 THEN 1 ELSE 0 END

...

Which promptly broke when I changed the column to a DATE. So let this be a warning: even something as innocuous as a DATETIME to DATE change needs QA and code search.

Laptopistan is getting crowded

This is purely anecdotal, of course, but on the occasions when I've ventured out from my home office to get some fresh air or avoid being pestered by my dog, I've seen more and more people working remotely. Your average Starbucks is completely overrun with laptop-facing contractors, remote workers, and students these days, as are most other public areas with seating and free WiFi. Getting a table all to yourself is becoming rare, and forget about finding an open power outlet!

This says to me that there is a serious hole in the co-working space market. Assuming these Starbucks campers order an average of a $4 latte per day (some get black coffee, which is cheaper, but some add a muffin), and the cost of a dedicated desk at Affinity Lab is on the order of $900, that works out to a price differential of $800/mo. Surely there must be a point somewhere in the middle for which one could add certain beneficial services while keeping the cost enough to attract customers.

Starbucks, after all, does not exactly inspire loyalty - the seating is functional, but rarely comfortable; there are never enough outlets; the coffee is okay but somewhat expensive; the WiFi is often slow; the environment is noisier than a worker would prefer; and the venues themselves hardly encourage hard work. If I could pay $200/mo for a similarly casual but better-outfitted space in which I could work, chat, and drink coffee, I would jump at the chance. I've been trying to check out DCIOLab, but they haven't emailed me back.

Makes me think there's a definite business opportunity here, but certainly a risky one. Timing is important - is NOW the inflection point in density of remote workers, or a year from now? You can quickly go broke renting a commercial space for a year with not enough customers. Is there enough consistency in the office support requirements of the proposed customers? A good coffee pot or two (or a Chemex or Aeropress, maybe) is definitely necessary, but a copier? Fax machine? Phones? A conference room?

These are ways to differentiate a workspace from a Starbucks, but I think they may raise the cost too much and get too little use. I don't need any of those things besides the coffee-making apparatus. In fact, all I want is coffee, power, internet, good light, and preferably high ceilings. Could one provide those things, pack people in as tightly as they do at Starbucks, and make money charging $200/mo?

Query plan troubleshooting

I was troubleshooting a medium-complexity query problem this morning in which the query execution within an app was orders of magnitude slower than the execution with the same parameters within Management Studio. I knew this could be due to different cached plans, but couldn't remember why.

Fortunately, Erland Sommerskog had written a detailed and helpful article explaining why this happens: http://www.sommarskog.se/query-plan-mysteries.html

And indeed, the app was using ODBC and therefore the results of sys.dm_exec_plan_attributes showed ARITHABORT OFF, while my instance of Management Studio had it set to ON, therefore generating a separate cached plan. I'm going to update my settings in SSMS to set this off by default, especially since it appears that it has no effect with ANSI_WARNINGS set to on, other than to foil attempts to troubleshoot slow procs.

Refreshing Intellisense in SQL Management Studio 2008

If you're like me, you create and drop a lot of tables while developing, and you may be getting annoyed with SQL Server Management Studio because it fills in the names of old tables while you're typing unless you hit Escape. Turns out that refreshing the Intellisense cache is as easy as

Ctrl-Shift-R