Estimating costs and credits for ArcGIS online subscription publishing

ArcMap 10.1 adds new capabilities for publishing map layers or entire compilations directly to the cloud using an ArcGIS Online (AGOL) Subscription.  AGOL subscriptions are a new type of GIS infrastructure that adds simplified cloud publishing to your ArcGIS desktop. Credits are the currency of subscriptions and each account comes with an annual amount of credits included. If you have administrator permissions in your AGOL subscription account, you can see your credits balance anytime as well as graphs of your credits usage by clicking the "View Status" link at the upper right of your "My Organization" page to get to the following screen, which will show you both a graph view and a timeline view of your credits usage:

d
Credits are used because some subscription tasks are cpu or bandwidth intensive and can't be supported in unlimited quantities. Credits are also used because AGOL subscriptions are a global product and outside the USA, each Esri distributor will set the subscription and credits fees in local currencies independently, based on local conditions. In the USA, credits cost 10 ¢ each, and the basic "Level 1" AGOL subscription includes 2500 credits per year. Additional credits can also be purchased in blocks of 1,000 for $100. Many subscription tasks are free or so cheap you won't use up your credits on them, but some tasks can be very expensive depending on how you set them up. This document is a guide to help you understand and manage expensive AGOL subscription tasks.


Credits Consumption for Different AGOL Tasks:

NO CHARGE: Any Data inbound/upload, any use of Esri Basemaps = no credits used

DATA:  1 GB data outbound from services, views, downloads = 6 credits  (60¢ per GB)

COMPUTATION:   1000 cache tiles created for Tiled Mapping service = 1 credit     (see table at right, for example: 500,000 tiles = 500 credits = $50)
Geocoding: 12.5 geocodes = 1 credit,    (10,000 geocodes = 800 credits = $80)

Tile Credit Costs

Tiles

Credits

Price

1million

1,000

$100

100,000

100

$10

10,000

10

$1

1,000

1

10¢

100

0.1


STORAGE: 1 GB file & Tiled Mapping storage = 1.2 credits/month,  14.4 credits/year    ($1.44 per GB per year)
10mb of Feature Access Service storage   = 2.4 credits/month, 28.8 credits/year     ($24 per GB per month, $288 per GB per year)

As you can see, there are 3 potentially expensive tasks: Geocoding, Feature Access service storage, and Tiled Mapping  tiles creation.  Currently, feature access services can burn credits at many times the posted rate above so it's wise to do a small test service of known size and let it run for a few days in order to get an accurate daily rate before you upload your main service. The first day of any service will always be pro-rated based on what time it spun up. The rest of this document will focus on map service tiles and how to estimate their costs.  The table at above right shows a more detailed breakdown of tiling costs for different quantities of tiles.

Lets take a look at tiling options in the typical ArcMap 10.1 cloud publishing workflow.  When you select the new "share as" option in the file menu of ArcMAP 10.1 you are given 2 choices, one of which is "Service".  On selecting service you can "Publish a service", which will then let you create a name for it and select a place to publish it.  If you have ArcServer set up at your site or in a cloud account you can add it here.  If you have an ArcGIS Online subscription you can add this subscription by providing your user ID and login, adding an option called "My Hosted Services"  to show up in your available publishing connections. 

d

 

You'll next get to describe the capabilities and options you want for your published map service

d

 

If you have checked "Tiled Mapping" then options for "Caching" and "Advanced Settings will appear:

d

Here you can select a tiling scheme.  We'll only cover the ArcGIS Online/Bing scheme as that is the standard for all ArcGIS online basemaps.  For this scheme you have 20 "levels of detail" to choose from, each of which can have radically different costs to create. Costs are based on the number of tiles, not their data volume, so to estimate this when all you have is your project area,  you need to know how much area is in each tile.

Tile Areas
The table below summarizes what these levels of detail actually represent in terms of scale, tiles and area, (demarcated every 5 levels for readability)
LOD:  Standard Level of Detail from 0 to 19
Tiles for Globe: Total number of tiles in each level covering the whole globe. (Formula is 2 to the power of (LOD*2), so each
  higher level has 2²=4 times the number of tiles as the preceding level, and every 5 levels has 4 to the 5th power or ≈1000 times.)
Scale: Nominal scale corresponding to that level of detail at 96dpi display standard
m-km per pixel: Width of a single pixel at the equator, in meters (m) or kilometers (k). Tile standard is 256x256 pixels, so this
   figure is obtained by dividing the equatorial length 40,075km by the pixel width of the map at that LOD, ie 256 pixels for the single tile at level 0.
Km per Tile Edge:  Linear width of a single tile, obtained as km per pixel * 256.
Area km² Per tile:  Area of a single tile,  square of  Km per Tile Edge.  (Can also be calculated by taking the total area of the map   and dividing it by the total number of tiles, but note that since the Mercator does NOT preserve area you cannot use the actual area of the earth, 510,072,000km2,  rather you must use the distorted area represented by the projection of 1,606,000,000km2. This figure also accounts for the map scheme stopping at 85º north and south so it doesn't go to infinity.)
Tiles per Area km²:  Inverse of previous

Tile Areas for Web Mercator Auxuilliary Sphere/Bing Scheme


LOD

Tiles for
Globe

Scale

m - km
Per pixel

Km per
Tile edge

Area km²
Per tile

Tiles per
Area km²

0

1

592m

156.5k

40,075

1,606m

 

1

4

296m

78.271k

20,037

401.5m

 

2

16

148m

39.136k

10,018

100m

 

3

64

74m

19.568k

5009

25m

 

4

256

37m

9.784k

2505

6.2m

 

5

1,024

18m

4.892k

1252

1.6m

 

6

4,096

9.2m

2.446k

626

400,000

0.000 0006

7

16,384

4.6m

1.223k

313

100,000

0.000 01

8

65,536

2.3m

611.5m

156

25,000

0.000 04

9

262,144

1.1m

305.7m

78

6,250

0.000 17

10

1,048,576

578k

152.9m

39

1,560

0.000 66

11

4,194,304

289k

76.44m

19.56

390

0.0026

12

16.7m

144k

38.22m

9.73

98

0.11

13

67.1m

72.2k

19.11m

4.89

24

0.43

14

268.4m

36k

9.55m

2.44

6

0.17

15

1,074m

18k

4.78m

1.22

.15

0.67

16

4,295m

9k

2.39m

0.61

.37

2.7

17

17,180m

4.5k

1.19m

0.305

.09

11

18

68,719m

2.25k

0.597m

0.153

.0234

43

19

274,878m

1.12k

0.298m

0.076

.0058

172

 

 

The next table shows how to translate your actual on-the ground area figures, assuming they are accurate.  Note the mention above about how Mercator is accurate for bearing and shape but distorts area, especially at high latitudes.  (Note how Antarctica and Greenland are exploded in the simple Mercator map at right. The shape of features and their directions to each other are correct, they are just many times larger at high latitudes)  To correct for what the Mercator projected area actually is on the ground we use what's called a "scale factor", a variable figure for each latitude that lets us convert actual ground area into area as it will appear in our map, therefore the basis for how tiles are generated. Between 0 and 30 degrees scale factor is essentially 1. It changes exponentially above 30 degrees.

Scale Factor

Latitude

Going from Ground to map,
Multiply area by:

Going from Map to Ground,
Multiply area by

0 equator

1

1

30º

1.15

0.87

45º

1.41

0.71

60º

2

0.5

85º

11.5

0.087

 

 

 

v

   

 

Tile Costs for different example areas
  The table below shows what it would cost to generate a Tiled Mapping service at each level of detail for 7 different example map projects or layers ranging in size from 10km² to the entire globe.  It was created by multiplying out the project area by the Tiles per Area km² column in the Tile Areas table for each LOD.    What is immediately apparent is for the lower LOD's anything that produces tiles in the 10,000 quantity range or less is essentially free, costing $1 or less. That covers over 2/3 of the options. However, many projects will instead want to answer the scale question more traditionally by using the maximum scale they can afford.  In that case, the highlighted figures represent what seems to be a useful threshold value of $1-$10. Any combination of area and detail above this threshold is less than $1 to generate the tiles.  Anything below it will cost more than $10 to generate the tiles.  Notice especially how fast costs go up as you go from layer to layer over that threshold. As the number of tiles increases four-fold with each successive layer, so do the costs.

Tile Costs for different example areas at different levels of detail

LOD

Scale

10km²

100km²

1,000km²

10,000km²

100,000km²

1million km²

Globe

LOD

0

592m

< $1

< $1

< $1

< $1

< $1

< $1

< $1

0

1

296m

< $1

< $1

< $1

< $1

< $1

< $1

< $1

1

2

148m

< $1

< $1

< $1

< $1

< $1

< $1

< $1

2

3

74m

< $1

< $1

< $1

< $1

< $1

< $1

< $1

3

4

37m

< $1

< $1

< $1

< $1

< $1

< $1

< $1

4

5

18m

< $1

< $1

< $1

< $1

< $1

< $1

< $1

5

6

9.2m

< $1

< $1

< $1

< $1

< $1

< $1

< $1

6

7

4.6m

< $1

< $1

< $1

< $1

< $1

< $1

16k  $1.6

7

8

2.3m

< $1

< $1

< $1

< $1

< $1

< $1

64k  $6

8

9

1.1m

< $1

< $1

< $1

< $1

< $1

< $1

273k  $27

9

10

578k

< $1

< $1

< $1

< $1

< $1

< $1

1m  $105

10

11

289k

< $1

< $1

< $1

< $1

< $1

< $1

4m  $417

11

12

144k

< $1

< $1

< $1

< $1

< $1

11,000  $1

16m  $1606

12

13

72.2k

< $1

< $1

< $1

< $1

< $1

43,000  $4.3

67m  $6745

13

14

36k

< $1

< $1

< $1

< $1

17,000  $1.7

170,000  $17

269m  $26,900

14

15

18k

< $1

< $1

< $1

< $1

67,000  $6.7

670,000  $67

1,076m  $107,602

15

16

9k

< $1

< $1

< $1

27,800  $2.8

300,000  $30

3m  $300

> $100k

16

17

4.5k

< $1

< $1

< $1

110,000  $11

1.1m  $110

11m  $1,100

> $100k

17

18

2.25k

< $1

< $1

< $1

430,000  $43

4.3m  $430

43m  $4,300

> $100k

18

19

1.12k

< $1

< $1

17,200 $1.7

1,720,000 $170

17m  $1,700

172m $17,200

> $100k

19