To challenge AWS' market domination, Google continues to beef up its cloud services and tools -- including those for management. Google Stackdriver, a hybrid cloud monitoring platform that works with both GCP and AWS, is one recent example of that.
Google acquired Stackdriver in 2014 to compete with Amazon Web Services tools, such as CloudWatch and CloudTrail. Today, Stackdriver is the monitoring, logging, diagnostics and troubleshooting foundation for Google Cloud Platform. Admins can use graphical summary dashboards, alerts, log analytics and debugging features for Google Compute Engine and App Engine, along with Amazon Elastic Compute Cloud (EC2) instances. Stackdriver also includes an extensive feature set that is customizable for a variety of management scenarios.
Key features of Google Stackdriver
Google Stackdriver ingests log data from any source via its published API. The tool collects data from GCP or EC2 instances, including Google containers. It automatically retains log data for 30 days, and then users can choose to archive that data in Google Cloud Storage.
Admins can browse, search and copy log data into BigQuery to enable SQL queries and analytics. Google Stackdriver also supports custom-defined log metrics that admins can export and visualize on the vendor's dashboard. Like AWS CloudTrail, Google Stackdriver provides an audit trail of privileged admin access to cloud instances and data sources.
Stackdriver users can create custom alert policies that trigger notifications whenever data falls outside preset boundaries or instances fail health checks or uptime criteria. Stackdriver can measure uptime as frequently as every minute with a customizable timeout value of up to 60 seconds. Uptime checks are summarized on a graphical dashboard that shows the status of various services in different GCP regions.
The tool can also use advanced probes to track activity on specific ports or HTTP host responses. The service supports several notification channels, including Slack, Campfire, HipChat and PagerDuty.
Stackdriver summarizes other types of data in charts and dashboards. Like traditional on-premises monitoring software, customized Stackdriver graphs can include a number of metrics, including both standard items, such as average CPU load or network statistics, as well as custom metrics, such as the number of API calls to a particular cloud-based application.
Google Stackdriver Trace is a distributed performance monitoring service that gathers and summarizes latency measurements from Google App Engine and Google Cloud load balancers. It can also work with any application using Trace's software development kit. Near-real-time data is displayed on the GCP Console and shows how long it takes applications to handle incoming URL requests or remote procedure calls to Google App Engine services. Users can also view the details of a complex request to see the latency and status codes for individual modules and identify the source of performance bottlenecks.
Google Stackdriver Debugger captures application state, such as local variables and the call stack, at any location in the source code without using explicit logging statements. Snapshots are displayed in a multipanel user interface that admins can share via its URL.
Google Stackdriver Pricing
Users can try Google Stackdriver for free but are limited to Google Cloud metrics and a 5 GB data repository. As of June 2017, full access costs $8 per resource, which is defined as an SQL, Compute or App Engine instance, per month with up to 10 GB of log data. Additional log data costs 50 cents per GB. For detailed cost estimates, see the online pricing calculator.
Usage and recommendations
Monitoring is fundamental to all cloud infrastructure management. So, if an enterprise chooses GCP, it should familiarize itself with Google Stackdriver to track resource usage, performance and uptime. Once admins navigate the basics, they should explore how to use Google Stackdriver in more sophisticated ways, such as:
- Testing disaster recovery plans: Use Google Stackdriver alerts to automate the deployment of backup VMs and other resources in response to error conditions or performance exceptions. Configure disaster recovery infrastructure with the GCP Deployment Manager.
- Automating load balancing: You can autoscale groups of GCP instances in response to Stackdriver metrics. While common scaling measures are CPU load or server capacity, with Stackdriver, users can trigger additional capacity in response to any custom parameter.
- Monitoring multicloud: Although Google Stackdriver doesn't monitor on-premises infrastructure, it does work with AWS and has over 500 predefined metrics that cover all of AWS' major services, including EC2, Elastic Block Store, CloudFront, Lambda, Relational Database Service and Redshift. Some AWS users might prefer Google Stackdriver over native AWS monitoring options, which are spread across multiple tools, such as CloudWatch, CloudTrail and X-Ray.
Explore Google cloud services and features
Familiarize yourself with Google cloud storage
Google unveils cloud partnership with SAP