You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
@@ -15,7 +15,7 @@ Cloud tiering, an optional feature of Azure File Sync, decreases the amount of l
15
15
16
16
When enabled, this feature stores only frequently accessed (hot) files on your local server. Infrequently accessed (cool) files are split into namespace (file and folder structure) and file content. The namespace is stored locally and the file content stored in an Azure file share in the cloud.
17
17
18
-
When a user opens a tiered file, Azure File Sync seamlessly recalls the file data from the file share in Azure.
18
+
When a user opens a tiered file, Azure File Sync seamlessly recalls the file data from the Azure file share.
19
19
20
20
## How cloud tiering works
21
21
@@ -31,36 +31,36 @@ For example, if your local disk capacity is 200 GiB and you want at least 40 GiB
31
31
32
32
#### Date policy
33
33
34
-
With the **date policy**, cool files are tiered to the cloud if they haven't been accessed (that is, read or written to) for x number of days. For example, if you noticed that files that have gone more than 15 days without being accessed are typically archival files, you should set your date policy to 15 days.
34
+
With the **date policy**, cool files are tiered to the cloud if they haven't been accessed (read or written to) for x number of days. For example, if you notice that files that have gone more than 15 days without being accessed are typically archival files, you should set your date policy to 15 days.
35
35
36
36
For more examples on how the date policy and volume free space policy work together, see [Choose Azure File Sync cloud tiering policies](file-sync-choose-cloud-tiering-policies.md).
37
37
38
38
### Windows Server data deduplication
39
39
40
-
Data deduplication is supported on volumes that have cloud tiering enabled beginning with Windows Server 2016. For more details, please see [Planning for an Azure File Sync deployment](file-sync-planning.md#data-deduplication).
40
+
Data deduplication is supported on volumes that have cloud tiering enabled beginning with Windows Server 2016. For details, see [Planning for an Azure File Sync deployment](file-sync-planning.md#data-deduplication).
41
41
42
42
### Cloud tiering heatmap
43
43
44
-
Azure File Sync monitors file access (read and write operations) over time and, based on how frequent and recent access is, assigns a heat score to every file. It uses these scores to build a "heatmap" of your namespace on each server endpoint. This heatmap is a list of all syncing files in a location with cloud tiering enabled, ordered by their heat score. Frequently accessed files that were recently opened are considered hot, while files that were barely touched and haven't been accessed for some time are considered cool.
44
+
Azure File Sync monitors file access (read and write operations) over time and assigns a heat score to every file based on how recently and frequently the file is accessed. It uses these scores to build a "heatmap" of your namespace on each server endpoint. This heatmap is a list of all syncing files in a location with cloud tiering enabled, ordered by their heat score. Frequently accessed files that were recently opened are considered hot, while files that were barely touched and haven't been accessed for some time are considered cool.
45
45
46
-
To determine the relative position of an individual file in that heatmap, the system uses the maximum of its timestamps, in the following order: MAX(Last Access Time, Last Modified Time, Creation Time).
46
+
To determine the relative position of an individual file in that heatmap, the system uses the maximum of its timestamps, in the following order: MAX(Last Access Time, Last Modified Time, Creation Time).
47
47
48
-
Typically, last access time is tracked and available. However, when a new server endpoint is created, with cloud tiering enabled, not enough time has passed to observe file access. If there is no valid last access time, the last modified time is used instead, to evaluate the relative position in the heatmap.
48
+
Typically, last access time is tracked and available. However, when a new server endpoint is created with cloud tiering enabled, not enough time has passed to observe file access. If there's no valid last access time, the last modified time is used instead, to evaluate the relative position in the heatmap.
49
49
50
-
The date policy works the same way. Without a last access time, the date policy will act on the last modified time. If that is unavailable, it will fall back to the create time of a file. Over time, the system will observe more file access requests and automatically start to use the self-tracked last access time.
50
+
The date policy works the same way. Without a last access time, the date policy will act on the last modified time. If that's unavailable, it will fall back to the create time of a file. Over time, the system will observe more file access requests and automatically start to use the self-tracked last access time.
51
51
52
52
> [!NOTE]
53
-
> Cloud tiering does not depend on the NTFS feature for tracking last access time. This NTFS feature is off by default and due to performance considerations, we do not recommend that you manually enable this feature. Cloud tiering tracks last access time separately.
53
+
> Cloud tiering does not depend on the NTFS feature for tracking last access time. This NTFS feature is off by default and due to performance considerations, we don't recommend that you manually enable this feature. Cloud tiering tracks last access time separately.
54
54
55
55
### Proactive recalling
56
56
57
57
When a file is created or modified, you can proactively recall a file to servers that you specify. Proactive recall makes the new or modified file readily available for consumption in each specified server.
58
58
59
-
For example, a globally distributed company has branch offices in the US and in India. In the morning (US time), information workers create a new folder and new files for a brand new project and work all day on it. Azure File Sync will sync folder and files to the Azure file share (cloud endpoint). Information workers in India will continue working on the project in their timezone. When they arrive in the morning, the local Azure File Sync enabled server in India needs to have these new files available locally, such that the India team can efficiently work off of a local cache. Enabling this mode prevents the initial file access to be slower because of on-demand recall and enables the server to proactively recall the files as soon as they were changed or created in the Azure file share.
59
+
For example, a globally distributed company has branch offices in the US and India. In the morning in the US, information workers create a new folder and files for a brand new project, and work all day on it. Azure File Sync will sync folder and files to the Azure file share (cloud endpoint). Information workers in India will continue working on the project in their time zone. When they arrive in the morning, the local Azure File Sync enabled server in India needs to have these new files available locally so the India team can efficiently work off of a local cache. Enabling this mode tells the server to proactively recall the files as soon as they're changed or created in the Azure file share, improving file access times.
60
60
61
-
If files recalled to the server are not needed locally, then the unnecessary recall can increase your egress traffic and costs. Therefore, only enable proactive recalling when you know that pre-populating a server's cache with recent changes from the cloud will have a positive effect on users or applications using the files on that server.
61
+
If files recalled to the server aren't needed locally, then the unnecessary recall can increase your egress traffic and costs. Therefore, only enable proactive recalling when you know that pre-populating a server's cache with recent changes from the cloud will have a positive effect on users or applications using the files on that server.
62
62
63
-
Enabling proactive recalling may also result in increased bandwidth usage on the server and may cause other relatively new content on the local server to be aggressively tiered due to the increase in files being recalled. In turn, tiering too soon may lead to more recalls if the files being tiered are considered hot by servers.
63
+
Enabling proactive recalling might also result in increased bandwidth usage on the server and could cause other relatively new content on the local server to be aggressively tiered due to the increase in files being recalled. In turn, tiering too soon might lead to more recalls if the files being tiered are considered hot by servers.
64
64
65
65
:::image type="content" source="media/storage-sync-files-deployment-guide/proactive-download.png" alt-text="An image showing the Azure file share download behavior for a server endpoint currently in effect and a button to open a menu that allows to change it.":::
66
66
@@ -72,57 +72,57 @@ Cloud tiering is the separation between namespace (the file and folder hierarchy
72
72
73
73
#### Tiered file
74
74
75
-
For tiered files, the size on disk is zero since the file content itself isn't being stored locally. When a file is tiered, the Azure File Sync file system filter (StorageSync.sys) replaces the file locally with a pointer (reparse point). The reparse point represents a URL to the file in the Azure file share. A tiered file has both the "offline" attribute and the FILE_ATTRIBUTE_RECALL_ON_DATA_ACCESS attribute set in NTFS so that third-party applications can securely identify tiered files.
75
+
For tiered files, the size on disk is zero because the file content itself isn't being stored locally. When a file is tiered, the Azure File Sync file system filter (StorageSync.sys) replaces the file locally with a pointer called a reparse point. The reparse point represents a URL to the file in the Azure file share. A tiered file has both the `offline` attribute and the `FILE_ATTRIBUTE_RECALL_ON_DATA_ACCESS` attribute set in NTFS so that third-party applications can securely identify tiered files.
76
76
77
77
78
78
79
79
#### Locally cached file
80
80
81
-
On the other hand, for a file stored in an on-premises file server, the size on disk is about equal to the logical size of the file since the entire file (file attributes + file content) is stored locally.
81
+
For files stored in an on-premises file server, the size on disk is about equal to the logical size of the file, because the entire file (file attributes + file content) is stored locally.
82
82
83
83
84
84
85
-
It's also possible for a file to be partially tiered (or partially recalled). In a partially tiered file, only part of the file is stored on disk. You may have partially recalled files on your volume if files are partially read by applications that support streaming access to files. Some examples are multimedia players and zip utilities. Azure File Sync is efficient and recalls only the requested information from the connected Azure file share.
85
+
It's also possible for a file to be partially tiered or partially recalled. In a partially tiered file, only part of the file is stored on disk. You might have partially recalled files on your volume if files are partially read by applications that support streaming access to files. Some examples are multimedia players and zip utilities. Azure File Sync is efficient and recalls only the requested information from the connected Azure file share.
86
86
87
87
> [!NOTE]
88
88
> Size represents the logical size of the file. Size on disk represents the physical size of the file stream that's stored on the disk.
89
89
90
90
## Low disk space mode
91
-
Disks with server endpoints can run out of space for several reasons, even with cloud tiering enabled. This could result in Azure File Sync not working as expected and even unusable. While it is not possible and not in control of Azure File Sync to prevent these occurrences completely, low disk space mode (new for Azure File Sync agent version 15.1) is designed to avoid a server endpoint reaching this situation.
91
+
Disks with server endpoints can run out of space for several reasons, even with cloud tiering enabled. This could result in Azure File Sync not working as expected and even becoming unusable. While it isn't possible for Azure File Sync to prevent these occurrences completely, low disk space mode (new for Azure File Sync agent version 15.1) is designed to avoid a server endpoint reaching this situation.
92
92
93
-
For server endpoints with cloud tiering enabled and volume free space policy set, if the free space on the volume reaches below the calculated threshold, then the volume is in low disk space mode.
93
+
For server endpoints with cloud tiering enabled and volume free space policy set, if the free space on the volume drops below the calculated threshold, then the volume is in low disk space mode.
94
94
95
95
In low disk space mode, the Azure File Sync agent does two things differently:
96
96
97
-
- Proactive Tiering: In this mode the File Sync agent tiers files proactively to the cloud . Sync agent checks for files to be tiered every minute instead of the normal frequency of every 1 hour. Volume free space policy tiering typically does not happen during initial upload sync until the full upload is complete, but in low disk space mode, tiering is enabled during the initial upload sync and files will be considered for tiering once the individual file has been uploaded to the Azure file share.
97
+
-**Proactive Tiering**: In this mode, the File Sync agent tiers files proactively to the cloud. The sync agent checks for files to be tiered every minute instead of the normal frequency of every hour. Volume free space policy tiering typically doesn't happen during initial upload sync until the full upload is complete; however, in low disk space mode, tiering is enabled during the initial upload sync, and files will be considered for tiering once the individual file has been uploaded to the Azure file share.
98
98
99
-
- Non-Persistent Recalls: When a user opens a tiered file, files recalled from the Azure File Share directly will not be persisted to the disk. Note that recalls initiated by the cmdlet Invoke-StorageSyncFileRecallare an exemption from this rule and will be persisted to disk.
99
+
-**Non-Persistent Recalls**: When a user opens a tiered file, files recalled from the Azure file share directly won't be persisted to the disk. Recalls initiated by the `Invoke-StorageSyncFileRecall` cmdlet are an exception to this rule and will be persisted to disk.
100
100
101
-
When the volume free space reaches above the threshold, Azure File Sync reverts to the normal state automatically. Note that low disk space mode only applies to servers with cloud tiering enabled and will always respect the volume free space policy.
101
+
When the volume free space surpasses the threshold, Azure File Sync reverts to the normal state automatically. Low disk space mode only applies to servers with cloud tiering enabled and will always respect the volume free space policy.
102
102
103
-
If a volume has two server endpoints, one with tiering-enabled and one without tiering, low disk space mode will only apply to the server endpoint where tiering is enabled.
103
+
If a volume has two server endpoints, one with tieringenabled and one without tiering, then low disk space mode will only apply to the server endpoint where tiering is enabled.
104
104
105
105
### How is the threshold for low disk space mode calculated?
106
-
The threshold is calculated by taking the minimum of the following three numbers:
106
+
Calculate the threshold by taking the minimum of the following three numbers:
107
107
- 10% of volume free space in GB
108
-
-Volume Free Space Policy in GB
109
-
-20 GB of volume free space
108
+
-Volume Free Space Policy in GB
109
+
-20 GB of volume free space
110
110
111
111
The following table includes some examples of how the threshold is calculated and when the volume will be in low disk space mode.
112
112
113
-
| Volume Size | Volume Free Space Policy | Current Volume Free Space | Threshold \= Min (10%, Volume Free Space Policy, 20GB) | Is Low Disk Space Mode? | Reason |
113
+
| Volume Size | Volume Free Space Policy | Current Volume Free Space | Threshold \= Min (10%, Volume Free Space Policy, 20 GB) | Is Low Disk Space Mode? | Reason |
|100GB | 7% (7GB) | 9% (9GB) | 7GB = Min (10GB, 7GB, 20GB) | No | Current Volume Free Space > Threshold |
116
-
|100GB | 7% (7GB) | 5% (5GB) | 7GB = Min (10GB, 7GB, 20GB) | Yes | Current Volume Free Space < Threshold |
117
-
|300GB | 8% (24GB) | 7% (21GB) | 20GB = Min (30GB, 24GB, 20GB) | No | Current Volume Free Space > Threshold |
118
-
|300GB | 8% (24GB) | 6% (18GB) | 20GB = Min (30GB, 24GB, 20GB) | Yes | Current Volume Free Space < Threshold |
115
+
|100 GB| 7% (7 GB) | 9% (9 GB) | 7GB = Min (10 GB, 7 GB, 20 GB) | No | Current Volume Free Space > Threshold |
116
+
|100 GB| 7% (7 GB) | 5% (5 GB) | 7GB = Min (10 GB, 7 GB, 20 GB) | Yes | Current Volume Free Space < Threshold |
117
+
|300 GB| 8% (24 GB) | 7% (21 GB) | 20GB = Min (30 GB, 24 GB, 20 GB) | No | Current Volume Free Space > Threshold |
118
+
|300 GB| 8% (24 GB) | 6% (18 GB) | 20GB = Min (30 GB, 24 GB, 20 GB) | Yes | Current Volume Free Space < Threshold |
119
119
120
120
121
121
### How does low disk space mode work with volume free space policy?
122
122
Low disk space mode always respects the volume free space policy. The threshold calculation is designed to make sure volume free space policy set by the user is respected.
123
123
124
124
### How to get out of low disk space mode?
125
-
Low disk space mode is designed to revert to normal behavior when volume free space is above the threshold. You can help speed up the process by looking for any recently created files outside the server endpoint location and moving them to a different disk if possible.
125
+
Low disk space mode is designed to revert to normal behavior when volume free space is above the threshold. You can help speed up the process by looking for any recently created files outside the server endpoint location and moving them to a different disk.
126
126
127
127
### How to check if a server is in Low Disk Space mode?
128
128
Event ID 19000 is logged to the Telemetry event log every minute for each server endpoint. Use this event to determine if the server endpoint is in low disk mode (IsLowDiskMode = true). The Telemetry event log is located in Event Viewer under Applications and Services\Microsoft\FileSync\Agent.
0 commit comments