Amazon Timestream

Timestream has no host or port to connect to — it is reached through the AWS SDK. AddisDB reuses the connection form for AWS details instead: the Host field becomes the region and the Username field becomes your AWS profile. There is no password, because authentication rides your existing AWS credential chain.

Who it is for

Timestream is fully managed and serverless: it tiers recent data into memory and older data into cheaper storage automatically, and you never size a cluster. Queries are ordinary SQL with time-series functions layered on.

The tiering is configuration, not magic — each table has a memory-store retention and a magnetic-store retention, and a query that reaches past the memory window is both slower and priced differently. Setting those two numbers to match how you actually query is most of the tuning there is.

It fits IoT fleets, application and infrastructure telemetry, and industrial data on AWS — particularly where the volume is large enough that manual storage tiering would be a chore.

Set up AWS access

1. Install the AWS CLI and run aws configure (or aws configure sso) to create a profile.
2. Verify it: aws timestream-write list-databases --region us-east-1.
3. The identity needs timestream:Select, timestream:DescribeEndpoints, and the List/Describe permissions for databases and tables.
4. Note the region your Timestream database lives in.
5. DescribeEndpoints is not optional — the SDK discovers the query endpoint through it, so a policy without it fails before any query runs.

Connect from AddisDB

1. New Connection → Amazon Timestream under Timeseries. The form changes shape: no port, no password, no SSL, no tunnel.
2. Put your AWS region in the AWS region field, for example us-east-1.
3. Put your profile name in AWS profile, or leave it blank to use the default credential chain.
4. Database is optional — set it to scope introspection to one Timestream database.
5. Test, then Save.

Querying the narrow model

Timestream stores measures in a narrow shape: each row is a timestamp, a set of dimensions, a measure name and a typed measure value. That is why queries carry the measure_value::double cast — you are naming which typed column to read.

SELECT bin(time, 5m) AS t, avg(measure_value::double) AS avg_temp
FROM "iot"."readings"
WHERE time > ago(6h)
GROUP BY bin(time, 5m)
ORDER BY t

ago() and bin() are the two functions you will use constantly: one bounds the range, the other buckets it. Keep the time predicate on every query — billing is by data scanned, and an unbounded range scans the whole retention period.

What AddisDB gives you

• Databases and tables listed in the sidebar.
• The Chart view — results carry a time column, so they render as a time series without extra work.
• A connection that is read-only by construction: the Query API only reads, and writes go through a separate ingestion API entirely.
• No stored credentials at all — authentication uses your AWS profile, SSO session or IAM role.
• Query tabs and history, notebooks, and ⌘K search across every connection.