Skip to content

Latest commit

 

History

History
154 lines (126 loc) · 10.1 KB

README.md

File metadata and controls

154 lines (126 loc) · 10.1 KB
Raw

Ingestly Endpoint

What's Ingestly?

Ingestly is a simple tool for ingesting beacons to Google BigQuery. Digital Marketers and Front-end Developers often want to measure user's activities on their service without limitations and/or sampling, in real-time, having ownership of data, within reasonable cost. There are huge variety of web analytics tools in the market but those tools are expensive, large footprint, less flexibility, fixed UI, and you will be forced to use SDKs including legacy technologies like document.write.

Ingestly is focusing on Data Ingestion from the front-end to Google BigQuery by leveraging Fastly's features. Also, Ingestly can be implemented seamlessly into your existing web site with in the same Fastly service, so you can own your analytics solution and ITP does not matter.

Ingestly provides:

  • Completely server-less. Fastly and Google manages all of your infrastructure for Ingestly. No maintenance resource required.
  • Near real-time data in Google BigQuery. You can get the latest data in less than seconds just after user's activity.
  • Fastest response time for beacons. The endpoint is Fastly's global edge nodes, no backend, response is HTTP 204 and SDK uses ASYNC request.
  • Direct ingestion into Google BigQuery. You don't need to configure any complicated integrations, no need to export/import by batches.
  • Easy to start. You can start using Ingestly within 2 minutes for free if you already have a trial account on Fastly and GCP.
  • WebKit's ITP friendly. The endpoint issues 1st party cookies with Secure and httpOnly flags.

Setup

You can use one of BigQuery and Elasticsearch, or both as a database for logging. Fastly support multiple log-streaming in the same configuration. BigQuery support SQL and faster query speed with massive logs. Elasticsearch supports super flexible schema-less data structure. If you are going to use custom data (*_attr variables) frequently, or you wish to utilize Kibana's great visualization features, Elasticearch is better choice. If you will get huge records from the giant website, or you wish to use Data Studio, BigQuery gives you better performance within reasonable cost.

Prerequisites

  • A Google Cloud Platform account, and a project used for Ingestly.
  • A Fastly account, and a service used for Ingestly.
  • This endpoint may use cookies named ingestlyId, ingestlySes and ingestlyConsent under your specified domain name.

Note that a GCP project and a Fastly service can be created for Ingestly or you can use your existing one.

Google Cloud Platform

Create a service account for Fastly

  1. Go to the GCP console, then open IAM & admin > service accounts.
  2. Create a service account like ingestly and grant a BigQuery > BigQuery Data Owner permission.
  3. Create a key and download it as JSON format.
  4. Open the JSON you just downloaded and note private_key and client_email.

Create a table for the log data on BigQuery

  1. Go to the GCP console, then open BigQuery.
  2. Create a dataset like Ingestly if you haven't had.
  3. Create a table with your preferred table name like access_log, then enable Edit as text in Schema section. (note your table name)
  4. Open BigQuery/table_schema file in this repository, copy the content and paste it to the schema text box of table creation modal.
  5. In the Partition and cluster settings section, Select timestamp column for partitioning.
  6. Specify action,category to the Clustering order (optional) field.
  7. Finish creating the table.

Elasticsearch

Create a user for Fastly

  1. Open Kibana UI.
  2. Go to Management > Security > Roles.
  3. Click top-right Create role button.
  4. Name this role as Ingestly
  5. Type ingestly-#{%F} into Index field manually. (an index name will be generated dynamically by strftime. in this case, an index is daily basis with YYYY-MM-DD formatted date.)
  6. Select create_index, create, index, read, write and monitor in Privileges field, then save.
  7. Go to Management > Security > Users
  8. Click top-right Create user button.
  9. Name this role as Ingestly and fill each field as you like.
  10. Select Ingestly from a role list, then save.

Put a mapping template to Elasticsearch

  1. Go to Dev Tools.
  2. Type PUT _template/ingestly into the first line of Dev Tools console.
  3. Open Elasticsearch/mapping_template.json file and copy & paste the content to the second line of Dev Tools console.
  4. Click the triangle icon on the first line (execute the command)

If you see Custom Analyzer related error message when you executed above process, you should choose one of the following selections.

A. Add Natural Language Analysis plugins to Elasticsearch. analysis-kuromoji and analysis-icu are recommended. B. Remove analysis section (from line 22 to line 40) from Elasticsearch/mapping_template.json to deactivate Analyzer.

Create an index pattern

  1. Go to Management > Kibana > Index Patterns.
  2. Click top-right Create index pattern button.
  3. Fill ingestly into Index Pattern field, then click Next step.
  4. Select timestamp from Time Filter field name pulldown, then click Create index pattern.

Fastly

Dictionaries

  1. Open Dictionaries under Data menu in CONFIGURE page under your service.
  2. Create a dictionary named ingestly_apikeys by clicking Create a dictionary button.
  3. Add an item with key as 2ee204330a7b2701a6bf413473fcc486, value as true from Add item link for ingestly_apikeys.
  4. In the same way, create a dictionary named ingestly_metadata by clicking Create a dictionary button.
  5. Add the following two items to the dictionary ingestly_metadata.
key value description
cookie_domain example.com A domain name of Cookies set by the Endpoint.
cookie_lifetime 31536000 A Cookie lifetime of Cookies set by the Endpoint.

Custom VCL

  1. Open Custom VCL in CONFIGURE page.
  2. Click Upload a VCL file button, then set preferred name like Ingestly, select ingestly.vcl and upload the file.

Integrate with Google BigQuery

  1. Open Logging in CONFIGURE page.
  2. Click CREATE ENDPOINT button and select Google BigQuery.
  3. Open attach a condition. link near highlighted CONDITION and select CREATE A NEW RESPONSE CONDITION.
  4. Enter a name like Data Ingestion and set (resp.status == 204 && req.url ~ "^/ingestly-ingest/(.*?)/\?.*" || resp.status == 200 && req.url ~ "^/ingestly-sync/(.*?)/\?.*") into Apply if… field.
  5. Fill information into fields:
    • Name : anything you want.
    • Log format : copy and paste the content of BigQuery/log_format file in this repository.
    • Email : a value from client_email field of GCP credential JSON file.
    • Secret key : a value from private_key field of GCP credential JSON file.
    • Project ID : your project ID of GCP.
    • Dataset : a dataset name you created for Ingestly. (e.g. Ingestly)
    • Table : a table name you created for Ingestly. (e.g. logs)
    • Template : this field can be empty but you can configure time-sliced tables if you enter like %Y%m%d.
  6. Click CREATE to finish the setup process.

Integrate with Elasticsearch

  1. Open Logging in CONFIGURE page.
  2. Click CREATE ENDPOINT button and select Elasticsearch.
  3. Open attach a condition. link near highlighted CONDITION and select CREATE A NEW RESPONSE CONDITION.
  4. Enter a name like Data Ingestion and set (resp.status == 204 && req.url ~ "^/ingestly-ingest/(.*?)/\?.*" || resp.status == 200 && req.url ~ "^/ingestly-sync/(.*?)/\?.*") into Apply if… field.
  5. Fill information into fields:
    • Name : anything you want.
    • Log format : copy and paste the content of Elasticsearch/log_format file in this repository.
    • URL : An endpoint URL of Elasticsearch cluster.
    • Index : An index name for Elasticsearch. Set ingestly.
    • BasicAuth user : An username for Elasticsearch authentication. Set Ingestly.
    • BasicAuth password : Set a password for user Ingestly on Elasticsearch cluster.
  6. Click CREATE to finish the setup process.

Integrate with Amazon S3

  1. Open Logging in CONFIGURE page.
  2. Click CREATE ENDPOINT button and select Amazon S3.
  3. Open attach a condition. link near highlighted CONDITION and select CREATE A NEW RESPONSE CONDITION.
  4. Enter a name like Data Ingestion and set (resp.status == 204 && req.url ~ "^/ingestly-ingest/(.*?)/\?.*" || resp.status == 200 && req.url ~ "^/ingestly-sync/(.*?)/\?.*") into Apply if… field.
  5. Fill information into fields:
    • Name : anything you want.
    • Log format : copy and paste the content of S3/log_format file in this repository. You can specify not only CSV but JSON format here ({ ... } form).
    • Timestamp format : (not necessary)
    • Bucket name : The name of the bucket in which to store the logs.
    • Access key : An access key of the service account that can write into the bucket above.
    • Secret key : An secret key of the service account that can write into the bucket above.
    • Period : Log rotation interval(seconds). e.g. 600 means 10 minutes.
    • Advanced options
      • Path : The path within the bucket for placing files. You may specify dynamic variables in strftime format. In order to use Athena's partitioning feature by date, the path name must include /date=%Y-%m-%d/ format.
      • Domain : The endpoint domain of your S3 bucket region (outside of US Standard region). e.g. Tokyo is s3.ap-northeast-1.amazonaws.com
      • Select a log line format : Blank. Otherwise the JSON format will be corrupted.
      • Gzip level : 9. The best compression to save the storage size.
  6. Click CREATE to finish the setup process.

Next Step