building a global multi-tenant monitoring platform
DESCRIPTION
Datapipe, an AWS Premier Consulting Partner, has built and customized a global monitoring platform specifically for AWS. This presentation discusses the challenges encountered when architecting this solution and provides a live demonstration of the platform and its specific monitoring capabilities.TRANSCRIPT
© 2014 Amazon.com, Inc. and its affiliates. All rights reserved. May not be copied, modified, or distributed in whole or in part without the express consent of Amazon.com, Inc.
Building a Global Multi-Tenant
Monitoring PlatformJeff Moody, Datapipe
#> whoami
• Jeff Moody– Cloud Engineer with Datapipe
– Principle engineer for architecting and building the Datapipe
Cloud Monitoring Platform product
– Extensive work in Cloud Computing and DevOps tool chains
Disclaimer
• My opinions are my own, not necessarily
representative of my employer
• Your Mileage May Vary
Who is Datapipe?
• Managed Service Provider– Headquartered in Jersey City, NJ
– Premier AWS Consulting Partner
– Single provider solution for mission critical IT
– Offices and data centers in New Jersey, Northern Virginia,
Silicon Valley, Iceland, London, Hong Kong, Shanghai, and
Singapore
– Fully and partially managed solutions running on Linux, UNIX,
and Windows
What are Datapipe’s Managed Service
Offerings?• Operating System Management & Support
• Application Stack Management & Support
• Network Management & Support
• DBA Management & Support
• Security & Compliance Management & Support– PCI, HIPAA
What does Datapipe bring to AWS?
• AWS Premier Consulting Partner– Exchange, MSP, Oracle, and SharePoint Competencies
• Managed AWS– Single Vendor Billing & Governance
– Cloud Provisioning of all AWS services including:EC2, S3, ELB, and CloudFront
– Infrastructure and Application Monitoring
– DBA Tuning and Optimization
– Managed Scaling
– Managed Security Groups
– 24x7 Issue Response and Resolution
What does Datapipe bring to AWS?
• AWS OptimizationGovernance, security, and financial optimization tools
– Cloud Reports
• Implement cloud budgeting and Reserved Instances across multiple
AWS accounts
• Push-button 24x7 response on critical security and utilization insights
– Datapipe Cloud Management Portal with Cloud Apps
• Single pane view for security, technology & billing
– Committed Instances
• The benefits of Reserved Instances without the upfront costs
What does Datapipe bring to AWS?
• AWS Direct Connect - Simplify Hybrid IT – Datapipe AWS Direct Connect Enabled Services
• Turnkey DX solutions - No CAPEX
• DBaaS (Oracle RAC, MS SQL Enterprise)
• Enterprise SANs, Crypto equipment, and more
– AWS Direct-Connect from:
• NJ to AWS US-East
• Virginia to AWS US-East
• Silicon Valley to AWS US-West (North California)
• London to AWS EU (Ireland)
• Singapore to AWS Asia Pacific (Singapore)
What is Cloud Reports?
Optimize and Forecast
Implement cloud budgeting with confidence and optimize Reserved Instances across multiple AWS
accounts and internal departments.
Insight Response and Resolution
Push-button 24x7 response on critical security and utilization insights from Datapipe’s award winning
global managed services support.
Custom Reporting for Individual Business Units
Dashboards and reports for the CEO, CTO, CFO, and development teams to get the most out of AWS.
PROBLEM: How do I efficiently use the vast amounts of operational data on my
AWS environments to provide insightful, actionable information?
ANSWER: Datapipe Cloud Reports for AWS
The Problem
• Late 2011
• Growing number of customers leveraging AWS
• Current monitoring platform built to work inside
datacenters
• Elasticity
The Solutions?
• Fix legacy monitoring platform– Non-starter as Datacenter monitoring moving to new SNMP-
based platform
– SNMP not the best choice for the Cloud
• Roll a new platform– Substantial development effort
– Support and maintenance nightmare
• Build a new platform from a different, non-SNMP vendor
The Solution
• Monitoring Vendor Provided Solution Challenges– Multi-tenancy
– Agents?
• Legacy Monitoring was Agent Based, caused a lot of finger-pointing
• Setup? (Automation toolset? Still often requires an agent)
– Centralized Management
• Easy deployment of “local” data collection to keep latency low
Datapipe Cloud Monitoring
• Powered by Zenoss Service Dynamics– Substantial number of customizations
– Agentless (leverages SSH and WMI)
– Scalable
– Supportable
Deployment Architecture
Master
Collector
Customer Server
Customer Server
Customer Server
Collector
Customer Server
Customer Server
Customer Server
Customer Server
Customer Server
Customer Server
Collector
VPC
Deployment Challenges
• AWS ZenPack requires one key per account for
auto discovery
• Distributing keys to all collectors is bad
• Programmability
Development Challenges
• Enrollment of Instance through a single-click– Update Security Group Rules
– Ensure presence of necessary keys on appropriate collector
– Ensure presence in Datapipe inventory
– Provide synchronization of information between monitoring,
inventory, and ticketing system
– Clean removal of instance on shutdown
Hybrid Data Solutions
• Datapipe already has collectors in every data
center– VLANs can be extended into the local collector for physical host
monitoring
– Single tool for full physical and virtual Cloud Monitoring
User Interface
• Zenoss UI is confusing
• Zenoss permissions are difficult to manage
• Abstracted into Datapipe Cloud Portal– One-click setup
What does Cloud Monitoring mean to me?
• Solution Escalation Action Plan (SEAP)– Alerts and procedures based on observed events
– i.e.: Scale an environment during a traffic spike, or who to
contact when an error is seen
• SLAs– Clear reporting to ensure SLAs are met
• Accountability & Peace of Mind– Understand the health of your environment
What does Cloud Monitoring Provide?
• Process monitoring– Auto-discovery for most major applications
• Web Stacks: Apache, nginx, HAProxy, Rails, Zope, IIS, Tomcat
• Databases: MySQL, MSSQL, PGSQL, Mongo, Riak
• URL Monitoring– HTTP content checks
– HTTP response codes
– Response times
• System monitoring– CPU, Memory, Disk
Instance Registration/Monitoring Demo
• This is a live demo
• Please be patient
Please give us your feedback on this
presentation
As a thank you, we will select prize
winners daily for completed surveys!