IT Infrastructure Through The Public Network Challenges And Solutions

IT Infrastructure Through the Public Network:Challenges & Solutions

$ whoami

• Martin Jackson – Uncommon Sense Consulting– Working in the IT Field since 1993– Linux and Virtualization Consultant specialising

in automated build and deployment of virtual infrastructures

– Infrastructure as Code Hacker– DevOps Advocate– Keen Judoka– @actionjack on Twitter– [email protected]

$ cat /infrastructure/info

Source: http://en.wikipedia.org/wiki/Cloud_computing

$ whatis iaas

• Outsourced Hardware• Outsourced Operating system• Outsourced Network• Self Managed• Typically available in Minutes• Pay per play

#1 Challenge:

Security

$ info security

• How do you protect your data in an infrastructure that you do not own or control?

$ cat security/access

• Protect your API keys and Use complex passwords– Cyber-Ark Enterprise Vault–Manage Engine Password Manager Pro – KeePass– APG and GPG

$ cat security/access

• Keep your systems patched (religiously)– Yum– Red Hat Network–Microsoft Update Network– Shavlik NetChk Protect– Apt

$ cat security/access

• Limit access to least privilege – Only create accounts for those who

“need” them– Create separate accounts per device– Do not allow direct access via privileged

user accounts e.g. Administrator or Root– Use audited privilege elevation e.g. sudo,

rootsh, sudosh, runas, shellrunas– Only use encrypted login mechanisms

e.g. ssh, ssl certificates

$ cat security/access

• Aggregate and monitor all login attempts– Splunk– Logstash– Graylog2– GFI Events Manager

$ cat security/data

• Encrypt your sensitive data before you place it into the cloud– PGP, GPG

• Keep it encrypted while in the cloud– TrueCrypt, LUKS

• Ensure encryption is maintained if data needs to be transmitted elsewhere– SCP, SSL, VPN, SSH

$ cat security/network

• If you need Secure Intra IaaS communication– SSL Auth– CohesiveFT’s VPN-Cubed– OpenVPN– Amazon Virtual Private Cloud

#2 Challenge:

Outages

$ whatis outage

• Unplanned unavailability of a service

"...in the cloud, you control your SLA..."

George Reese, founder enStratus Networks LLC

$ whatis outage

“large-scale, essentially self-managed and commoditised infrastructure-as-a-service (IaaS) has price benefits but, if

things go wrong, they do so in a big way”

Dr Aydin Kurt-Elli, Lumison

$ whatis outage

• Vendor: TerremarkOutage Date: March 17, 2010Outage Duration: 7 hoursReason for Outage: Terremark's vCloud Express services suffered an outage after a bout of connectivity loss in its Miami data center. T he outage resulted in intermittent periods of connectivity with high data packet loss starting at 11:54 a.m. eastern and lasting more than seven hours, ending at 7:05 p.m. eastern time. According to Apparent Networks' Cloud Performance Center, during the outage access to systems in Terremark's Miami data center was severely degraded and often unavailable, affecting many businesses using Terremark's vCloud Express services.Severity: Medium

• http://www.crn.com/slide-shows/applications-os/225701829/10-biggest-cloud-outages-of-2010-so-far.htm;jsessionid=o+AywGYF+Mv5w3ZoWChIbQ**.ecappj01?pgno=5

$ whatis outage

• Vendor:Rackspace• Outage Date:2011-02-01• Outage Duration:30 minutes• Reason for Outage:DNS Issue Causes MySQL Server

Outage.An unspecified DNS issue prevented users from connecting to MySQL and making external API calls. Rackspace resolved the issue and advised their users to refresh their browsers to view the site properly.

• Severity:Low • http://outagecenter.com/rackspace-cloud-reports/

cloud-sites-dfw1-wc2-degraded-2/

$ whatis outage

• Vendor:Rackspace• Outage Date:April 28,2011• Outage Duration:6 hours• Reason for Outage:At approximately 4:00 PM (CDT)

customers began to experience connectivity issues related to Domain Name System (DNS) on Jungle Disk/Cloud Drive.The issue was identified to be an error with hostname translations on a single DNS server. This server was returning erroneous DNS information.an emergency maintenance to change the DNS configuration was performed In order to mitigate the issue.

• Severity: Medium  

$ whatis outage

• Vendor: Amazon Web ServiceOutage Date: April 21, 2011Outage Duration: UnknownReason for Outage: Amazon began reporting trouble on its Service Health Dashboard about 5 a.m. Eastern today. At 5:16 a.m., the site reported connectivity issues that were affecting its Relational Database Service, which is used to manage a relational database in the cloud, across multiple zones in the eastern U.S. A networking event early this morning triggered a large amount of re-mirroring of EBS volumes in US-EAST-1.The re-mirroring created a shortage of capacity in one of the US-EAST-1 Availability Zones, which impacted new EBS volume creation as well as the pace with which we could re-mirror and recover affected EBS volumes. Amazon also reported problems with its EC2, or Elastic Compute Cloud, a service that provides pay-as-you-go compute capacity in the cloud. The company also reported issues with its EBS, or Elastic Block Storage, which is storage related to the EC2 service.

• Severity: High• http://

www.computerworld.com/s/article/9216064/Amazon_gets_black_eye_from_cloud_outage

$ whatis outage

• Vendor: Amazon Web ServiceOutage Date: August 08, 2011Outage Duration: 30 MinutesReason for Outage: The issue happened in the networks that connect the Availability Zones to the internet. The event began when a southern router inside one of Availability Zones briefly stopped exchanging route information with all adjacent devices, going into an incommunicative state. Upon re-establishing its health, the router began advertising an unusable route to other southern routers in other Availability Zones, deviating from its configuration and bypassing the standard protocol restriction on how routes are allowed to flow. The bad default internet route was picked up and used by the routers in other Availability Zones. Internet traffic from multiple Availability Zones in US East was immediately not routable out to the internet through the border. The issue was resolved by removing the router from service.Severity:Medium

• http://outagecenter.com/category/amazon-web-services-reports/amazon-elastic-compute-cloud-ec2-north-virginia/

$ whatis outage

• Failure is the new black, expect it and embrace it

• Design for failure and build your infrastructures to be redundant on 5 different levels– Physical– Virtual resource– Availability zone– Region– Cloud

#3 Challenge:

Standards

$ find standard

• Cloud standards and Interoperability• To be honest they don’t exist yet…

http://www.infoq.com/articles/problem-with-cloud-computing-standardization

$ cat standard/api

• Many different clouds…• Many ways to interact with them…• All do the same sort thing…• Let abstract them– Deltacloud– Libcloud– Jclouds

$ cat standard/api/deltacloud

• http://incubator.apache.org/deltacloud/

• Ruby client

http://www.infoq.com/articles/problem-with-cloud-computing-standardization

require 'deltacloud'

api_url = 'http://localhost:3001/api' api_name = 'mockuser' api_password = 'mockpassword' client = DeltaCloud.new( api_name, api_password, api_url )

$ cat standard/api/libcloud

• http://libcloud.apache.org/• Python client

http://www.infoq.com/articles/problem-with-cloud-computing-standardization

from libcloud.compute.types import Providerfrom libcloud.compute.providers import get_driver

EC2_ACCESS_ID = 'your access id'EC2_SECRET_KEY = 'your secret key'

Driver = get_driver(Provider.EC2)conn = Driver(EC2_ACCESS_ID, EC2_SECRET_KEY)

$ cat standard/api/jclouds

• http://libcloud.apache.org/• Java client

http://www.infoq.com/articles/problem-with-cloud-computing-standardization

ComputeServiceContext context = new ComputeServiceContextFactory().createContext("aws-ec2", accesskeyid, secretkey, ImmutableSet.<Module> of(new Log4JLoggingModule(), new JschSshClientModule()));

#4 Challenge:

Monitoring and Management

$ service monitor status

• Pay per play monitoring or fixed instance

• On premise or Off• Ramping up and tearing down of

instances• Focus on Service monitoring vs host

monitoring• Monitoring tool must have an api

$ service monitor status

• Next Generation Cloud Monitoring Services

• Cloudkick - https://www.cloudkick.com

• Pingdom - http://www.pingdom.com• Watchmouse -

http://www.watchmouse.com

• Monitis – http://www.monitis.com

$ service management status

• Provision within minutes – Ready in Days???

• If it takes 5 minutes to get a Virtual Machine

• How long are you willing to wait to use it?• Data Center Automation Tools can help– Puppet– Chef– CFEngine

$ cat management/puppet

• http://puppetlabs.com/

package { 'openssh-server':ensure => installed,}

$ cat management/chef

• http://www.opscode.com/

package "openssh-server" do action :installend

$ cat management/chef

• http://cfengine.com/

control: any:: actionsequence = ( packages )

DefaultPkgMgr = ( rpm ) RPMcommand = ( /bin/rpm ) RPMInstallCommand = ( "/usr/bin/yum -y install %s" )

packages: any:: openssh-server action=install

#5 Challenge:

Governance

$ make governance

• The game has changed and you’ll need to change with it

• Conway's law applies:“...organizations which design systems ... are constrained to produce designs

which are copies of the communication structures of these organizations.”

Challenge:

Questions

$ cat links

• http://www.accenture.com/us-en/outlook/Pages/outlook-online-2011-challenges-cloud-computing.aspx

• http://www.infoq.com/articles/problem-with-cloud-computing-standardization

• http://www.computerworld.com/s/article/9217158/Cloud_interoperability_Problems_and_best_practices

• http://www.theaccidentalsuccessfulcio.com/cloud-computing/cio-cloud-computing-101-problems-with-clouds

• http://nylawblog.typepad.com/suigeneris/2009/11/does-cloudcomputing-compromise-clients.html

• http://horicky.blogspot.com/2009/08/multi-tenancy-in-cloud-computing.html

• http://www.cio.com/article/488478/The_Trouble_with_Cloud_Vendor_Lock_in

• http://www.agathongroup.com/blog/2010/04/cloud-computing-and-latency/