Microsoft SQL Server 2008 R2 Specifications download pdf (Page 177)

Application Development CHAPTER 8 157

As you might guess, the key to working with windows is to have a clear understanding of

the time span that each window covers. There are three types of window streams that Stream-

Insight supports—hopping windows, snapshot windows, and count windows. In a hopping

windows stream, each window spans an equal time period. In a snapshot windows stream,

the size of a window depends on the events that it contains. By contrast, the size of a count

windows stream is not xed, but varies according to a specied number of consecutive event

start times.

To create a hopping window, you specify both the time span that the window covers (also

known as window size) and the time span between the start of one window and the start of

the next window (also known as hop size). For example, assume that you need to create win-

dows that cover a period of one hour, and a new window starts every 15 minutes, as shown in

Figure 8-3. In this case, the window size is one hour and the hop size is 15 minutes. Here is the

code to create a hopping windows stream and count the events in each window:

var outputStream = from eventWindow in

inputStream.HoppingWindow(TimeSpan.FromHours(1), TimeSpan.FromMinutes(15))

select new { count = eventWindow.Count() };

60 90 120

Input

events

Time (minutes)

Hopping

windows

FIGURE 8-3 Hopping windows

When there are no gaps and there is no overlap between the windows in the stream,

hopping windows are also called tumbling windows. Figure 8-2, shown earlier, provides an

example of tumbling windows. The window size and hop size are the same in a tumbling

1 2 ... 172 173 174 175 176 177 178 179 180 181 182 ... 235 236

No comments

Microsoft SQL Server 2008 R2 Specifications Page 177