Data migration route
Amazon S3 to Server drives migration
Amazon S3 is cloud file storage, and Server drives are plain files on equipment the firm controls. The folders move as they are, and the work is in permissions, paths and links.
Amazon S3Cloud- Documents: Carried across
- Emails: Carried with mapping
Server drivesFiles- Custom fields: Kept as a readable archive
- 1 carried
- 1 mapped
- 1 archived
What moves
What moves from Amazon S3 to Server drives
Each row is a kind of record Amazon S3 holds, whether Server drives have a place for it, and what is planned. On this route one kind of record carries across, one kind needs mapping and one kind has no place in Server drives and is archived.
| Record | What is planned | ||
|---|---|---|---|
| DocumentsFiles stored in or linked from the system | HeldObjects in buckets; a folder is only the part of the key before a slash. | Has a placeFiles in folders, named and dated during the load. | Carried across |
| EmailsEmails saved to matters or to the store | Held in partOnly where messages were saved to the bucket as files such as .msg or .eml. | Has a place in partAs .msg or .eml files in each matter’s correspondence folder. | Carried with mapping |
| Custom fieldsFields and record types the firm added itself | Held in partObjects can carry user-defined metadata and tags as name and value pairs. | No place | Kept as a readable archive |
What the plans mean
- Carried across
- Server drives has a place for this kind of record. It is loaded there and checked against Amazon S3.
- Carried with mapping
- Server drives holds this differently, or in part. Each field or value is matched to its place before loading, and you see the mapping first.
- Kept as a readable archive
- Server drives has no place for it. It is exported in a format that opens without Amazon S3 and handed to the firm as a read-only archive.
The table states what the two products hold and what NewLaw plans for. It is not a promise that every record moves: that is settled against your own data in the analysis.
This kind of move
Same folders, new home.
Copying folders from Amazon S3 to Server drives sounds like a drag and drop, and for a few gigabytes it is. For a firm’s whole document store the trouble is in the detail. The two stores have different limits on path length and different rules about characters in file names. Files open on someone’s desk during the copy are skipped. Modified dates are reset unless the copy is made with a tool that preserves them. Each of these is checked on a trial copy first.
Permissions and links do not travel. Who can see which folder is set up again in Server drives, and it is a good moment to replace access granted to individuals over the years with groups. Shared links sent to clients stop working when the old store is switched off, so the live ones are listed and reissued. Documents that point at other documents by path, common in spreadsheets, are found and reported. The final pass copies only what changed since the trial.
Amazon S3 is hosted by its maker and Server drives sit on equipment the firm controls. The data leaves a service that was backed up for the firm and arrives somewhere the firm must back up itself, so backup and access are arranged before the export is copied in. What Amazon S3 will release depends on its export tools, and that is confirmed in the analysis.
Where the data is now
Leaving Amazon S3
Moving a firm’s files out of S3 starts with finding out who put them there and how they are named. A bucket filled by a person usually mirrors a client and matter folder tree. One filled by an application may hold nothing but numbered objects, with the real file names kept in that application’s database. Those are different jobs, and the analysis says which one this is.
Folders are part of the name
S3 has no folders. A key such as Clients/Nguyen/2019-0412/Letter.pdf is one long name, and the console draws folders from the slashes. Keys can run to 1,024 bytes, far longer than Windows or SharePoint accept, and may contain characters they reject. Those keys are listed in the analysis and shortened by an agreed rule.
Modified dates are upload dates
The date S3 records against an object is the date it was written to the bucket. A letter drafted in 2014 and uploaded in 2021 shows as 2021. Unless the original date was kept in object metadata or in an index, it cannot be recovered from S3, and document dates in the new system will reflect that.
Archive classes need a restore
Objects in the S3 Glacier Flexible Retrieval and Deep Archive storage classes cannot be read straight away. Each has to be restored to a temporary copy first, which takes time and is charged by AWS. Closed-matter archives are often stored this way, so the restore is requested before the trial load, not during it.
Versions exist only if enabled
Versioning is off unless someone switched it on for the bucket. Where it is on, every overwrite keeps the earlier object in full, and a deleted file sits behind a delete marker and is not removed. The firm decides whether earlier versions and deleted objects move, because a plain copy of the bucket brings only the current ones.
Application buckets need the database
Where a practice or document system used the bucket as its store, the keys are often identifiers and not file names. The matter, the document title and the author live in that system’s database. Copying the bucket alone produces thousands of unnamed files, so the objects are matched to their records before anything is loaded.
Where it is going
Arriving in Server drives
Some firms leave a practice or document system and keep their documents on a server drive, either as the working store or as the archive of a system being switched off. The files will arrive without the database that described them. Everything the firm wants to keep knowing about a document has to be put into its path, its name or an index file.
Agree the tree and the names
A fixed pattern is chosen, for example client name and number, then matter number and short description, then a standard set of subfolders. Document titles from the old system become file names. Where two documents share a title, a date or sequence number is added so nothing is overwritten.
Budget the path length
Share name, client, matter, subfolder and file name all count towards the 260 character limit that many Windows programs still apply. The pattern is tested against the firm’s longest client names and document titles before the load. Descriptions are trimmed by rule, never by cutting off the file extension.
Dates have to be set deliberately
A file written to the server today is dated today unless its modified date is set from the old system’s record. Document dates are how lawyers find things in a folder. Which date each file is to carry is agreed in the analysis, and the trial load checks a sample.
Emails become message files
Emails filed to matters in the old system are written out as .msg or .eml files in a correspondence folder, named by date, sender and subject so they sort in order. Attachments stay inside the message. Subjects with characters Windows forbids are cleaned, and long ones cut.
Keep an index beside the files
A spreadsheet or small database listing every document with its matter, author, original date, document type and new path is the only place the old system’s profile data survives. It lets someone find a document by something other than its name. What the index records, and who keeps it, is decided in the analysis.
How NewLaw runs it
Four steps, in this order.
Nothing financial lands in Server drives on this route, which makes it repeatable: a load can be run again until it is right.
Analysis
We read what Amazon S3 holds and how it is organised, then agree with you where each kind of record will sit in Server drives. Anything with no place there stays behind or is archived, and you see which before the move.
Clean-up
Duplicate clients, matters and folders are merged where they can be identified, and naming is put into one format, before anything is copied. It is not always possible to find every one, so some may carry across.
Trial load
A trial is loaded into Server drives and checked against Amazon S3: record and file counts, the links between records, document dates, and a sample opened by your own people. Exceptions are listed for you to decide.
Cutover
There is no financial cutover on this route. After a short freeze on changes in Amazon S3, a final pass over a weekend picks up what changed since the trial. Amazon S3 is kept read only as the fall-back record.
Fixed price, agreed after the analysis. How NewLaw moves legal data
Start
Start this migration
Tell us what is in Amazon S3 and roughly how much of it. The first step is the analysis, and the price is fixed once it is done.
Step 1 of 5
What happens after you send it
- A free first call
We go through your answers with you and ask about anything they leave open.
- An analysis of what Amazon S3 holds
What is in it, how its records link, and where each piece would sit in Server drives. Anything with no place there comes across as a note or stays behind, and you see which before the move.
- A fixed price in writing
Set after the analysis, so it reflects what is in your data. You have it before any work starts.
Beyond the data
Structure, permissions and handover
Server drives are storage, not an application, so there is no product to configure. What is agreed and set up is the folder structure, who can open what, and how the result is handed to the firm.
A server drive does only what the firm and its IT provider make it do. Setting one up as a document store means building the share, deciding who can open which folders, and arranging the protections a document system would have supplied: backups, earlier versions and a way to recover a deleted folder. The filing rule is written down and taught.
What is agreed and set up
- Share and top-level folders for clients, precedents and administration
- Security groups and folder permissions, with restricted matters separate
- Drive letter mapped for every user at sign-in
- Backup schedule and shadow copies for earlier versions
- Written naming rule for client, matter and document names
- Read-only area for closed matters and archived system exports
Questions
Amazon S3 to Server drives: what firms ask
Our IT provider set the bucket up. What do you need from them?
Access to read the bucket, arranged in whatever way the firm and its provider decide, and the region it sits in. They are also asked whether versioning is on, whether any objects are in an archive storage class, and whether a person or an application wrote the files. Moving the files out means reading them, so write access is not part of the request.
Will our client and matter folders look the same in the new system?
In outline, if the keys follow a client and matter pattern. Each part of the key between slashes becomes a folder, or is read as the client, the matter number and the document name where the destination files by matter. Keys that fit no pattern are listed for the firm to place by hand or leave in an unsorted folder.
Can the bucket stay as our archive after the move?
It can. A firm may move its active matters into the new system and leave closed files where they are. What matters then is that someone at the firm, not only the IT provider, holds the AWS account details and knows how to find one closed file by client name. That is worth writing down before the project ends.
Do documents from Amazon S3 keep their folders and dates in Server drives?
Server drives store documents. Files in folders, named and dated during the load. Each file is placed under the client and matter it came from, and created and modified dates are carried, so a document still shows when it was written. The trial load is checked by count and by opening a sample.
What happens to records Server drives have no place for?
On this route that is custom fields. They are exported from Amazon S3 in formats that open without it, spreadsheets and PDFs, and handed to the firm as a read-only archive. You see the list before the move.
What has to be decided before Amazon S3 data is loaded into Server drives?
Emails are carried with mapping, so each mapping is agreed with you first. The firm also decides which closed matters and how many versions of a document are in scope.
Other routes
Other routes from Amazon S3 and into Server drives
Where to start
Before and after the move.
Choosing the system comes before a migration. Setting it up and putting AI to work come after.
Choose your systems
Technology stack selection
Decide which practice management system your firm runs, with the evidence and the cost beside it.
About four weeks for a firm of about 15 people. Fixed price, set after a free first call.
Move your data
Data migration
Your data is cleaned up before anything moves, then a trial load is checked against the old system.
Fixed price, agreed after the analysis.
Set up your new system
Implementation
Your new system set up around the way your firm works: configuration, document automation and training.
A small firm of 1 to 10 users typically takes 4 to 8 weeks. A mid-sized firm of 10 to 50 users takes 8 to 16 weeks.
Put AI to work
AI advisory and enablement
A small group of your people learn to use an approved AI tool on real work, under written rules.
Two to four weeks for a small first group. Fixed price, set after a free first call.
Amazon S3 and Server drives are the names and marks of their owners, used here only to identify the two systems.
Moving from Amazon S3 to Server drives?
Tell us what you are moving. You get a plan and a fixed price after the analysis.
We reply within one business day.