Question

以下查询在较高的缩放级别（边界框小于2度经度，纬度为0.5度）时表现非常好，但随着边界框变大而显着降低。该表包含7~800万行文本和位置数据存储作为几何列中的点。

我尝试过不同的网格配置，但是当边界框比@ north = 41.123029000000002，@ east = -72.935406，@ south = 40.296503999999999，@ west = -75.077740000000006

更大时，性能总是会下降

有什么想法吗？谢谢〜马特

我在不同的缩放级别中包含了一个表格。

declare @filter geometry
select @filter = GEOMETRY::STGeomFromText(
                            'LINESTRING(' + CONVERT(varchar,@west) + ' ' + CONVERT(varchar,@south) + ',' + CONVERT(varchar,@east) + ' ' + CONVERT(varchar,@north) + ')'
                            ,4326
                            ).STEnvelope();

select
    x.Id
    ,x.[Timestamp]
    ,x.Location
    ,x.[Text]
from (
    select top(1000)
        t.Id
        ,t.[Timestamp]
        ,t.Location
        ,t.[Text]
    from dbo.Table1 AS t with (nolock, index([SPIX_Table1_Location_HIGH]))
    inner join containstable(
                dbo.Table1
                ,[text]
                , N'FORMSOF(INFLECTIONAL, "word1") | FORMSOF(INFLECTIONAL, "word2") | FORMSOF(INFLECTIONAL, "wordN")'
        ) as r on t.Id = r.[KEY]
    where t.Location.Filter(@filter) = 1
        AND t.[Hour] >= @HourId
    order by r.[RANK] desc
) as x
order by x.[Timestamp] desc
option (maxdop 1)

缩放等级2 = @ north = 74.542308000000006，@ east = 94.21875，@ south = -24.370607，@ west = -180 缩放等级20 = @ north = 40.711250999999997，@ east = -74.006050000000002，@ south = 40.710847000000001，@ west = -74.007096000000004

这是空间索引：

CREATE SPATIAL INDEX [SPIX_Table1_location_HIGH] ON [dbo].[Table1] 
(
    [location]
)USING  GEOMETRY_GRID 
WITH (
BOUNDING_BOX =(-180, -90, 180, 90), GRIDS =(LEVEL_1 = HIGH,LEVEL_2 = HIGH,LEVEL_3 = HIGH,LEVEL_4 = HIGH), 
CELLS_PER_OBJECT = 16, FILLFACTOR = 70) ON [PRIMARY]
GO

按缩放级别请求持续时间（1000次随机测试）：

Level   Duration0to5    Duration5to10   Duration10to15  Duration15to20  DurationGreaterThan20
2   0   0   0   0   26
3   0   0   0   0   42
4   0   0   0   0   57
5   0   0   0   0   60
6   0   0   0   0   54
7   0   0   0   1   65
8   0   2   5   6   34
9   0   3   7   10  6
10  5   23  25  14  1
11  13  26  18  3   0
12  17  31  7   0   0
13  48  11  0   0   0
14  48  6   0   0   0
15  47  1   0   0   0
16  57  0   0   0   0
17  48  8   0   0   0
18  44  3   0   0   0
19  63  5   0   0   0
20  47  3   0   0   0
ALL 437 122 62  34  345

典型查询的行计数和执行时间，w / out top（1000），强制空间索引：

if (@level=2)
select @north=74.542308000000006,@east=94.21875,@south=-24.370607,@west=-180 --70,404 rows, 2 minutes w/ spatial
if (@level=3)
select @north=61.978465999999997,@east=-5.451886,@south=9.7052770000000006,@west=-142.56126 -- 57,911 rows, 1m22s w/ spatial
if (@level=4)
select @north=52.614061999999997,@east=-39.729228999999997,@south=26.230861999999998,@west=-108.283917 -- 45,636 rows, 1m23s w/ spatial
if (@level=5)
select @north=46.992624999999997,@east=-56.867901000000003,@south=33.775959999999998,@west=-91.145245000000003 -- 32,386 rows, 26s w/ spatial
if (@level=6)
select @north=43.934699999999999,@east=-65.437236999999996,@south=37.323439,@west=-82.575908999999996 -- 19,998 rows, 13s w/ spatial
if (@level=7)
select @north=42.343530999999999,@east=-69.721905000000007,@south=39.037540999999997,@west=-78.291240999999999 -- 11,256 rows, 13s w/ spatial
if (@level=8)
select @north=41.532438999999997,@east=-71.864238999999998,@south=39.879399999999997,@west=-76.148906999999994 -- 6,147 rows, 4s w/ spatial
if (@level=9)
select @north=41.123029000000002,@east=-72.935406,@south=40.296503999999999,@west=-75.077740000000006 -- 3,667 rows, 3s w/ spatial

SQL Server空间查询在低缩放级别下执行效果不佳

0 个答案: