Mercurial > hg > orthanc-book
annotate Sphinx/source/plugins/object-storage.rst @ 500:4481882d9c83
typo
author | Alain Mazy <alain@mazy.be> |
---|---|
date | Tue, 01 Sep 2020 18:22:01 +0200 |
parents | d255e02eb89d |
children | 4e426dec4fee |
rev | line source |
---|---|
451 | 1 .. _object-storage: |
2 | |
3 | |
4 Cloud Object Storage plugins | |
5 ============================ | |
6 | |
7 .. contents:: | |
8 | |
499
d255e02eb89d
updated object-storage doc for 1.0.0
Alain Mazy <alain@mazy.be>
parents:
464
diff
changeset
|
9 Release notes |
d255e02eb89d
updated object-storage doc for 1.0.0
Alain Mazy <alain@mazy.be>
parents:
464
diff
changeset
|
10 ------------- |
d255e02eb89d
updated object-storage doc for 1.0.0
Alain Mazy <alain@mazy.be>
parents:
464
diff
changeset
|
11 |
d255e02eb89d
updated object-storage doc for 1.0.0
Alain Mazy <alain@mazy.be>
parents:
464
diff
changeset
|
12 Release notes are available `here |
d255e02eb89d
updated object-storage doc for 1.0.0
Alain Mazy <alain@mazy.be>
parents:
464
diff
changeset
|
13 <https://hg.orthanc-server.com/orthanc-object-storage/file/default/NEWS>`__ |
451 | 14 |
15 Introduction | |
16 ------------ | |
17 | |
18 Osimis freely provides the `source code | |
19 <https://hg.orthanc-server.com/orthanc-object-storage/file/default/>`__ of 3 plugins | |
20 to store the Orthanc files in `Object Storage <https://en.wikipedia.org/wiki/Object_storage>`__ | |
21 at the 3 main providers: `AWS <https://aws.amazon.com/s3/>`__, | |
22 `Azure <https://azure.microsoft.com/en-us/services/storage/blobs/>`__ & | |
23 `Google Cloud <https://cloud.google.com/storage>`__ | |
24 | |
25 Storing Orthanc files in object storage and your index SQL in a | |
26 managed database allows you to have a stateless Orthanc that does | |
27 not store any data in its local file system which is highly recommended | |
28 when deploying an application in the cloud. | |
29 | |
30 | |
459
a4ed4e883337
highlighting the pre-compiled binaries for google, aws and azure
Sebastien Jodogne <s.jodogne@gmail.com>
parents:
453
diff
changeset
|
31 Pre-compiled binaries |
a4ed4e883337
highlighting the pre-compiled binaries for google, aws and azure
Sebastien Jodogne <s.jodogne@gmail.com>
parents:
453
diff
changeset
|
32 --------------------- |
a4ed4e883337
highlighting the pre-compiled binaries for google, aws and azure
Sebastien Jodogne <s.jodogne@gmail.com>
parents:
453
diff
changeset
|
33 |
a4ed4e883337
highlighting the pre-compiled binaries for google, aws and azure
Sebastien Jodogne <s.jodogne@gmail.com>
parents:
453
diff
changeset
|
34 These plugins are used to interface Orthanc with commercial and |
a4ed4e883337
highlighting the pre-compiled binaries for google, aws and azure
Sebastien Jodogne <s.jodogne@gmail.com>
parents:
453
diff
changeset
|
35 proprietary cloud services that you accept to pay. As a consequence, |
a4ed4e883337
highlighting the pre-compiled binaries for google, aws and azure
Sebastien Jodogne <s.jodogne@gmail.com>
parents:
453
diff
changeset
|
36 the Orthanc project doesn't freely provide pre-compiled binaries for |
a4ed4e883337
highlighting the pre-compiled binaries for google, aws and azure
Sebastien Jodogne <s.jodogne@gmail.com>
parents:
453
diff
changeset
|
37 Docker, Windows, Linux or OS X. These pre-compiled binaries do exist, |
a4ed4e883337
highlighting the pre-compiled binaries for google, aws and azure
Sebastien Jodogne <s.jodogne@gmail.com>
parents:
453
diff
changeset
|
38 but are reserved to the companies who have subscribed to a |
a4ed4e883337
highlighting the pre-compiled binaries for google, aws and azure
Sebastien Jodogne <s.jodogne@gmail.com>
parents:
453
diff
changeset
|
39 `professional support contract |
a4ed4e883337
highlighting the pre-compiled binaries for google, aws and azure
Sebastien Jodogne <s.jodogne@gmail.com>
parents:
453
diff
changeset
|
40 <https://www.osimis.io/en/services.html#cloud-plugins>`__ by |
a4ed4e883337
highlighting the pre-compiled binaries for google, aws and azure
Sebastien Jodogne <s.jodogne@gmail.com>
parents:
453
diff
changeset
|
41 Osimis. Although you are obviously free to compile these plugins by |
a4ed4e883337
highlighting the pre-compiled binaries for google, aws and azure
Sebastien Jodogne <s.jodogne@gmail.com>
parents:
453
diff
changeset
|
42 yourself (instructions are given below), purchasing such support |
a4ed4e883337
highlighting the pre-compiled binaries for google, aws and azure
Sebastien Jodogne <s.jodogne@gmail.com>
parents:
453
diff
changeset
|
43 contracts makes the Orthanc project sustainable in the long term, to |
a4ed4e883337
highlighting the pre-compiled binaries for google, aws and azure
Sebastien Jodogne <s.jodogne@gmail.com>
parents:
453
diff
changeset
|
44 the benefit of the worldwide community of medical imaging. |
a4ed4e883337
highlighting the pre-compiled binaries for google, aws and azure
Sebastien Jodogne <s.jodogne@gmail.com>
parents:
453
diff
changeset
|
45 |
a4ed4e883337
highlighting the pre-compiled binaries for google, aws and azure
Sebastien Jodogne <s.jodogne@gmail.com>
parents:
453
diff
changeset
|
46 |
451 | 47 Compilation |
48 ----------- | |
49 | |
50 .. highlight:: text | |
51 | |
52 The procedure to compile the plugins is quite similar of that for the | |
53 :ref:`core of Orthanc <compiling>` although they usually require | |
54 some prerequisites. The documented procedure has been tested only | |
55 on a Debian Buster machine. | |
56 | |
57 The compilation of each plugin produces a shared library that contains | |
58 the plugin. | |
59 | |
60 | |
61 AWS S3 plugin | |
62 ^^^^^^^^^^^^^ | |
63 | |
64 Prerequisites: Compile the AWS C++ SDK:: | |
65 | |
66 $ mkdir ~/aws | |
67 $ cd ~/aws | |
68 $ git clone https://github.com/aws/aws-sdk-cpp.git | |
69 $ | |
70 $ mkdir -p ~/aws/builds/aws-sdk-cpp | |
71 $ cd ~/aws/builds/aws-sdk-cpp | |
72 $ cmake -DBUILD_ONLY="s3;transfer" ~/aws/aws-sdk-cpp | |
73 $ make -j 4 | |
74 $ make install | |
75 | |
76 Prerequisites: Install `vcpkg <https://github.com/Microsoft/vcpkg>`__ dependencies:: | |
77 | |
78 $ ./vcpkg install cryptopp | |
79 | |
80 Compile:: | |
81 | |
82 $ mkdir -p build/aws | |
83 $ cd build/aws | |
84 $ cmake -DCMAKE_TOOLCHAIN_FILE=[vcpkg root]\scripts\buildsystems\vcpkg.cmake ../../orthanc-object-storage/Aws | |
85 | |
86 Azure Blob Storage plugin | |
87 ^^^^^^^^^^^^^^^^^^^^^^^^^ | |
88 | |
89 Prerequisites: Install `vcpkg <https://github.com/Microsoft/vcpkg>`__ dependencies:: | |
90 | |
91 $ ./vcpkg install cpprestsdk | |
92 | |
93 | |
94 Compile:: | |
95 | |
96 $ mkdir -p build/azure | |
97 $ cd build/azure | |
98 $ cmake -DCMAKE_TOOLCHAIN_FILE=[vcpkg root]\scripts\buildsystems\vcpkg.cmake ../../orthanc-object-storage/Azure | |
99 | |
100 Google Storage plugin | |
101 ^^^^^^^^^^^^^^^^^^^^^ | |
102 | |
103 Prerequisites: Install `vcpkg <https://github.com/Microsoft/vcpkg>`__ dependencies:: | |
104 | |
105 $ ./vcpkg install google-cloud-cpp | |
106 $ ./vcpkg install cryptopp | |
107 | |
108 Compile:: | |
109 | |
110 $ mkdir -p build/google | |
111 $ cd build/google | |
112 $ cmake -DCMAKE_TOOLCHAIN_FILE=[vcpkg root]\scripts\buildsystems\vcpkg.cmake ../../orthanc-object-storage/google | |
113 | |
114 | |
115 Configuration | |
116 ------------- | |
117 | |
118 .. highlight:: json | |
119 | |
120 AWS S3 plugin | |
121 ^^^^^^^^^^^^^ | |
122 | |
123 Sample configuration:: | |
124 | |
125 "AwsS3Storage" : { | |
126 "BucketName": "test-orthanc-s3-plugin", | |
127 "Region" : "eu-central-1", | |
128 "AccessKey" : "AKXXX", | |
463 | 129 "SecretKey" : "RhYYYY", |
499
d255e02eb89d
updated object-storage doc for 1.0.0
Alain Mazy <alain@mazy.be>
parents:
464
diff
changeset
|
130 "Endpoint": "", // custom endpoint |
d255e02eb89d
updated object-storage doc for 1.0.0
Alain Mazy <alain@mazy.be>
parents:
464
diff
changeset
|
131 "ConnectionTimeout": 30, // connection timeout in seconds |
d255e02eb89d
updated object-storage doc for 1.0.0
Alain Mazy <alain@mazy.be>
parents:
464
diff
changeset
|
132 "RequestTimeout": 1200, // request timeout in seconds (max time to upload/download a file) |
d255e02eb89d
updated object-storage doc for 1.0.0
Alain Mazy <alain@mazy.be>
parents:
464
diff
changeset
|
133 "MigrationFromFileSystemEnabled": false, // see below |
d255e02eb89d
updated object-storage doc for 1.0.0
Alain Mazy <alain@mazy.be>
parents:
464
diff
changeset
|
134 "StorageStructure": "flat" // see below |
451 | 135 } |
136 | |
464 | 137 The **EndPoint** configuration is used when accessing an S3 compatible cloud provider. I.e. here is a configuration to store data on Scaleway:: |
138 | |
139 "AwsS3Storage" : { | |
140 "BucketName": "test-orthanc", | |
141 "Region": "fr-par", | |
142 "AccessKey": "XXX", | |
143 "SecretKey": "YYY", | |
144 "Endpoint": "s3.fr-par.scw.cloud" | |
145 }, | |
146 | |
451 | 147 Azure Blob Storage plugin |
148 ^^^^^^^^^^^^^^^^^^^^^^^^^ | |
149 | |
150 Sample configuration:: | |
151 | |
152 "AzureBlobStorage" : { | |
153 "ConnectionString": "DefaultEndpointsProtocol=https;AccountName=xxxxxxxxx;AccountKey=yyyyyyyy===;EndpointSuffix=core.windows.net", | |
499
d255e02eb89d
updated object-storage doc for 1.0.0
Alain Mazy <alain@mazy.be>
parents:
464
diff
changeset
|
154 "ContainerName" : "test-orthanc-storage-plugin", |
d255e02eb89d
updated object-storage doc for 1.0.0
Alain Mazy <alain@mazy.be>
parents:
464
diff
changeset
|
155 "MigrationFromFileSystemEnabled": false, // see below |
d255e02eb89d
updated object-storage doc for 1.0.0
Alain Mazy <alain@mazy.be>
parents:
464
diff
changeset
|
156 "StorageStructure": "flat" // see below |
451 | 157 } |
158 | |
159 | |
160 Google Storage plugin | |
161 ^^^^^^^^^^^^^^^^^^^^^ | |
162 | |
163 Sample configuration:: | |
164 | |
165 "GoogleCloudStorage" : { | |
166 "ServiceAccountFile": "/path/to/googleServiceAccountFile.json", | |
499
d255e02eb89d
updated object-storage doc for 1.0.0
Alain Mazy <alain@mazy.be>
parents:
464
diff
changeset
|
167 "BucketName": "test-orthanc-storage-plugin", |
d255e02eb89d
updated object-storage doc for 1.0.0
Alain Mazy <alain@mazy.be>
parents:
464
diff
changeset
|
168 "MigrationFromFileSystemEnabled": false, // see below |
d255e02eb89d
updated object-storage doc for 1.0.0
Alain Mazy <alain@mazy.be>
parents:
464
diff
changeset
|
169 "StorageStructure": "flat" // see below |
451 | 170 } |
171 | |
172 | |
499
d255e02eb89d
updated object-storage doc for 1.0.0
Alain Mazy <alain@mazy.be>
parents:
464
diff
changeset
|
173 Migration & Storage structure |
d255e02eb89d
updated object-storage doc for 1.0.0
Alain Mazy <alain@mazy.be>
parents:
464
diff
changeset
|
174 ----------------------------- |
d255e02eb89d
updated object-storage doc for 1.0.0
Alain Mazy <alain@mazy.be>
parents:
464
diff
changeset
|
175 |
d255e02eb89d
updated object-storage doc for 1.0.0
Alain Mazy <alain@mazy.be>
parents:
464
diff
changeset
|
176 The **StorageStructure** configuration allows you to select the way objects are organized |
d255e02eb89d
updated object-storage doc for 1.0.0
Alain Mazy <alain@mazy.be>
parents:
464
diff
changeset
|
177 within the storage (``flat`` or ``legacy``). |
500 | 178 Unlike the traditional file system in which Orthanc uses 2 levels |
499
d255e02eb89d
updated object-storage doc for 1.0.0
Alain Mazy <alain@mazy.be>
parents:
464
diff
changeset
|
179 of folders, object storages usually have no limit on the number of files per folder and |
d255e02eb89d
updated object-storage doc for 1.0.0
Alain Mazy <alain@mazy.be>
parents:
464
diff
changeset
|
180 therefore all objects are stored at the root level of the object storage. This is the |
d255e02eb89d
updated object-storage doc for 1.0.0
Alain Mazy <alain@mazy.be>
parents:
464
diff
changeset
|
181 default ``flat`` behaviour. Note that, in the ``flat`` mode, an extension `.dcm` or `.json` |
d255e02eb89d
updated object-storage doc for 1.0.0
Alain Mazy <alain@mazy.be>
parents:
464
diff
changeset
|
182 is added to the filename which is not the case in the legacy mode. |
d255e02eb89d
updated object-storage doc for 1.0.0
Alain Mazy <alain@mazy.be>
parents:
464
diff
changeset
|
183 |
500 | 184 The ``legacy`` behaviour mimics the Orthanc File System convention. This is actually helpful |
499
d255e02eb89d
updated object-storage doc for 1.0.0
Alain Mazy <alain@mazy.be>
parents:
464
diff
changeset
|
185 when migrating your data from a file system to an object storage since you can copy all the file |
d255e02eb89d
updated object-storage doc for 1.0.0
Alain Mazy <alain@mazy.be>
parents:
464
diff
changeset
|
186 hierarchy as is. |
d255e02eb89d
updated object-storage doc for 1.0.0
Alain Mazy <alain@mazy.be>
parents:
464
diff
changeset
|
187 |
d255e02eb89d
updated object-storage doc for 1.0.0
Alain Mazy <alain@mazy.be>
parents:
464
diff
changeset
|
188 Note that you can not change this configuration once you've uploaded the first files in Orthanc. |
d255e02eb89d
updated object-storage doc for 1.0.0
Alain Mazy <alain@mazy.be>
parents:
464
diff
changeset
|
189 |
d255e02eb89d
updated object-storage doc for 1.0.0
Alain Mazy <alain@mazy.be>
parents:
464
diff
changeset
|
190 The **MigrationFromFileSystemEnabled** configuration has been for Orthanc to continue working |
d255e02eb89d
updated object-storage doc for 1.0.0
Alain Mazy <alain@mazy.be>
parents:
464
diff
changeset
|
191 while you're migrating your data from the file system to the object storage. While this option is enabled, |
d255e02eb89d
updated object-storage doc for 1.0.0
Alain Mazy <alain@mazy.be>
parents:
464
diff
changeset
|
192 Orthanc will store all new files into the object storage but will try to read/delete files |
d255e02eb89d
updated object-storage doc for 1.0.0
Alain Mazy <alain@mazy.be>
parents:
464
diff
changeset
|
193 from both the file system and the object storage. |
d255e02eb89d
updated object-storage doc for 1.0.0
Alain Mazy <alain@mazy.be>
parents:
464
diff
changeset
|
194 |
d255e02eb89d
updated object-storage doc for 1.0.0
Alain Mazy <alain@mazy.be>
parents:
464
diff
changeset
|
195 This option can be disabled as soon as all files have been copied from the file system to the |
d255e02eb89d
updated object-storage doc for 1.0.0
Alain Mazy <alain@mazy.be>
parents:
464
diff
changeset
|
196 object storage. Note that Orthanc is not copying the files from one storage to the other; you'll |
d255e02eb89d
updated object-storage doc for 1.0.0
Alain Mazy <alain@mazy.be>
parents:
464
diff
changeset
|
197 have to use a standard ``sync`` command from the object-storage provider. |
d255e02eb89d
updated object-storage doc for 1.0.0
Alain Mazy <alain@mazy.be>
parents:
464
diff
changeset
|
198 |
d255e02eb89d
updated object-storage doc for 1.0.0
Alain Mazy <alain@mazy.be>
parents:
464
diff
changeset
|
199 |
452 | 200 Sample setups |
201 ------------- | |
202 | |
203 You'll find sample deployments and more info in the `Orthanc Setup Samples repository <https://bitbucket.org/osimis/orthanc-setup-samples/src/master/#markdown-header-for-osimisorthanc-pro-image-users>`__ . | |
204 | |
205 | |
451 | 206 Client-side encryption |
207 ---------------------- | |
208 | |
209 Although all cloud providers already provide encryption at rest, the plugins provide | |
210 an optional layer of client-side encryption . It is very important that you understand | |
211 the scope and benefits of this additional layer of encryption. | |
212 | |
213 Rationale | |
214 ^^^^^^^^^ | |
215 | |
216 Encryption at rest provided by cloud providers basically compares with a file-system disk encryption. | |
217 If someone has access to the disk, he won't have access to your data without the encryption key. | |
218 | |
219 With cloud encryption at rest only, if someone has access to the "api-key" of your storage or if one | |
220 of your admin inadvertently make your storage public, `PHI <https://en.wikipedia.org/wiki/Protected_health_information>`__ will leak. | |
221 | |
222 Once you use client-side encryption, you'll basically store packets of meaningless bytes on the cloud infrastructure. | |
223 So, if an "api-key" leaks or if the storage is misconfigured, packets of bytes will leak but not PHI since | |
224 no one will be able to decrypt them. | |
225 | |
226 Another advantage is that these packets of bytes might eventually not be considered as PHI anymore and eventually | |
227 help you meet your local regulations (Please check your local regulations). | |
228 | |
229 However, note that, if you're running entirely in a cloud environment, your decryption keys will still | |
230 be stored on the cloud infrastructure (VM disks - process RAM) and an attacker could still eventually gain access to this keys. | |
231 | |
232 If Orthanc is running in your infrastructure with the Index DB on your infrastructure, and files are store in the cloud, | |
233 the master keys will remain on your infrastructure only and there's no way the data stored in the cloud could be decrypted outside your infrastructure. | |
234 | |
235 Also note that, although the cloud providers also provide client-side encryption, we, as an open-source project, | |
236 wanted to provide our own implementation on which you'll have full control and extension capabilities. | |
237 This also allows us to implement the same logic on all cloud providers. | |
238 | |
239 Our encryption is based on well-known standards (see below). Since it is documented and the source code is open-source, | |
240 feel-free to have your security expert review it before using it in a production environment. | |
241 | |
242 Technical details | |
243 ^^^^^^^^^^^^^^^^^ | |
244 | |
245 Orthanc saves 2 kind of files: DICOM files and JSON summaries of DICOM files. Both files contain PHI. | |
246 | |
452 | 247 When configuring the plugin, you'll have to provide a **Master Key** that we can also call the **Key Encryption Key (KEK)**. |
451 | 248 |
452 | 249 For each file being saved, the plugin will generate a new **Data Encryption Key (DEK)**. This DEK, encrypted with the KEK will be pre-pended to the file. |
451 | 250 |
251 If, at any point, your KEK leaks or you want to rotate your KEKs, you'll be able to use a new one to encrypt new files that are being added | |
252 and still use the old ones to decrypt data. You could then eventually start a side script to remove usages of the leaked/obsolete KEKs. | |
253 | |
254 To summarize: | |
255 | |
452 | 256 - We use `Crypto++ <https://www.cryptopp.com/>`__ to perform all encryptions. |
451 | 257 - All keys (KEK and DEK) are AES-256 keys. |
258 - DEKs and IVs are encrypted by KEK using CTR block cipher using a null IV. | |
259 - data is encrypted by DEK using GCM block cipher that will also perform integrity check on the whole file. | |
260 | |
261 The format of data stored on disk is therefore the following: | |
262 | |
263 - **VERSION HEADER**: 2 bytes: identify the structure of the following data currently `A1` | |
264 - **MASTER KEY ID**: 4 bytes: a numerical ID of the KEK that was used to encrypt the DEK | |
265 - **EIV**: 32 bytes: IV used by DEK for data encryption; encrypted by KEK | |
266 - **EDEK**: 32 bytes: the DEK encrypted by the KEK. | |
267 - **CIPHER TEXT**: variable length: the DICOM/JSON file encrypted by the DEK | |
268 - **TAG**: 16 bytes: integrity check performed on the whole encrypted file (including header, master key id, EIV and EDEK) | |
269 | |
270 Configuration | |
271 ^^^^^^^^^^^^^ | |
272 | |
273 .. highlight:: text | |
274 | |
275 AES Keys shall be 32 bytes long (256 bits) and encoded in base64. Here's a sample OpenSSL command to generate such a key:: | |
276 | |
277 openssl rand -base64 -out /tmp/test.key 32 | |
278 | |
279 Each key must have a unique id that is a uint32 number. | |
280 | |
281 .. highlight:: json | |
282 | |
283 Here's a sample configuration file of the `StorageEncryption` section of the plugins:: | |
284 | |
285 { | |
499
d255e02eb89d
updated object-storage doc for 1.0.0
Alain Mazy <alain@mazy.be>
parents:
464
diff
changeset
|
286 "GoogleCloudStorage" : { |
d255e02eb89d
updated object-storage doc for 1.0.0
Alain Mazy <alain@mazy.be>
parents:
464
diff
changeset
|
287 "StorageEncryption" : { |
d255e02eb89d
updated object-storage doc for 1.0.0
Alain Mazy <alain@mazy.be>
parents:
464
diff
changeset
|
288 "Enable": true, |
d255e02eb89d
updated object-storage doc for 1.0.0
Alain Mazy <alain@mazy.be>
parents:
464
diff
changeset
|
289 "MasterKey": [3, "/path/to/master.key"], // key id - path to the base64 encoded key |
d255e02eb89d
updated object-storage doc for 1.0.0
Alain Mazy <alain@mazy.be>
parents:
464
diff
changeset
|
290 "PreviousMasterKeys" : [ |
d255e02eb89d
updated object-storage doc for 1.0.0
Alain Mazy <alain@mazy.be>
parents:
464
diff
changeset
|
291 [1, "/path/to/previous1.key"], |
d255e02eb89d
updated object-storage doc for 1.0.0
Alain Mazy <alain@mazy.be>
parents:
464
diff
changeset
|
292 [2, "/path/to/previous2.key"] |
d255e02eb89d
updated object-storage doc for 1.0.0
Alain Mazy <alain@mazy.be>
parents:
464
diff
changeset
|
293 ], |
d255e02eb89d
updated object-storage doc for 1.0.0
Alain Mazy <alain@mazy.be>
parents:
464
diff
changeset
|
294 "MaxConcurrentInputSize" : 1024 // size in MB |
d255e02eb89d
updated object-storage doc for 1.0.0
Alain Mazy <alain@mazy.be>
parents:
464
diff
changeset
|
295 } |
451 | 296 } |
297 } | |
298 | |
299 **MaxConcurrentInputSize**: Since the memory used during encryption/decryption can grow up to a bit more | |
300 than 2 times the input, we want to limit the number of threads doing concurrent processing according | |
301 to the available memory instead of the number of concurrent threads. Therefore, if you're currently | |
302 ingesting small files, you can have a lot of thread working together while, if you're ingesting large | |
303 files, threads might have to wait before receiving a "slot" to access the encryption module. |