SlideShare a Scribd company logo
1 of 7
Download to read offline
Copyright © 2016, Adobe
Adobe Campaign Page 1
1.
2.
3.
Hardware Sizing Guide
Overview
This document provides general guidelines for Adobe Campaign Version 6 deployment at your on-premise data center or virtualized
cloud environment. This type of deployment, referred to as “hybrid” or “mid-sourcing”, places the Campaign marketing instance and
marketing database under your operational control, while using Adobe Cloud Messaging services to send emails, SMS and/or SMPP
messages, and collect email open, bounce and click tracking data.
The marketing instance is the portion of the Adobe Campaign architecture that drives all marketing activity, and stores all recipient data
and analytic data returned by campaigns. The marketing instance is created by a set of on-premise servers running Adobe Campaign
services, and a relational database.
The recommendations in this document do not apply if you are using a fully-hosted Adobe Campaign instance (deployed in Adobe Cloud
Services), nor do they apply to Adobe Campaign Standard edition.
Scenarios
Deployment diagrams and hardware sizing recommendations are provided for three representative scenarios:
Moderate Size – 1 million active recipients in the system,
Large Size – 10 million active recipients in the system, and
Enterprise – 25 million active recipients, with additional transactional messaging.
Assumptions
We also assume the following types of usage for all three scenarios:
Large email campaigns are sent twice each week, to approximately 50% of your active recipients
Direct mailings are generated once per month for each recipient in the system
SMS messages are sent to approximately 10% of your active recipients each month
The database schema defining each recipient has been extended with one additional table, containing about two hundred bytes
of data for each recipient
Adobe Campaign Interactions are used to add offers to outgoing email
Email tracking data is retained in the Campaign system for 90 days
Compatibility Matrix
See https://docs.campaign.adobe.com/doc/AC6.1/en/technicalResources/matrix.html
General Guidelines
Campaign is a database-centric application, and database server performance is critical. Running workflows, segmentation, tracking data
uploads, inbound Interactions, analytics and other activities all generate database activity. In general, the size and frequency of these
operations determine the size of your database server(s).
The application servers in your marketing instance require enough CPU and memory to run workflows and respond to SOAP API calls,
including requests from Campaign Console users. CPU requirements can be significant for workflows that use outbound Interactions with
complex offer rules, workflows that execute custom Javascript, and web applications with high traffic levels.
Document Purpose
General guidelines for estimating hardware requirements for on-premise Campaign deployment. This document is also
attached as a PDF, which may be shared with customers.
Copyright © 2016, Adobe
Adobe Campaign Page 2
Campaign web applications may also be deployed on the marketing instance app servers, or on separate web server systems. Since
web application workloads conflict with critical workflows and Campaign Console users, web applications and inbound Interactions may
be deployed to separate servers, to ensure that core Campaign functionality runs reliably with good performance.
Reverse Proxies
The Campaign architecture enforces high security by using SSL over HTTP (HTTPS) to communicate between your marketing instance
and Adobe Cloud Messaging. Security, reliability and availability are enforced by using reverse proxies in a “demilitarized zone” (DMZ)
subnet to isolate and secure the marketing instance servers and database.
When deploying proxies on your own systems, the following hardware recommendation is sufficient for all scenarios described in this
document:
2G + dual-core CPU,hz 4 GB RAM, RAID 1, 2 x 80GB HDD
Notes: Adobe Campaign includes a reverse proxy which may be embedded into IIS or Apache; the hardware recommendation above is
applicable only if you decide to go with the Adobe Campaign reverse proxy.
The Application Server proxy is set up in an active/passive configuration, with HTTPS terminated at the proxy. Adobe will provide you
with the exclusive list of URLs that should be relayed to the Adobe Campaign server in your deployment environment.
Scenario 1: Moderate-Size Deployment
Volume Estimates
Channel Comments
Active recipients 1 million
Email 4.2 million/month
Direct Mail 1 million/month
Mobile SMS 100,000/month
Peak daily email volume 500,000
For these volumes, a pair of Adobe Campaign application server systems provide all functionality for Adobe Campaign Client users,
Campaign web applications and workflow execution. For 1 million active recipients and this email volume, application server workloads
are not CPU or I/O intensive; most of the stress is on the database.
Your web servers are shown in the secure zone, showing how recipient clicks on links in email come back to your system. In fact, these
links get redirected through Adobe Cloud Messaging, so they can be tracked.
Application Servers
Copyright © 2016, Adobe
Adobe Campaign Page 3
In this scenario, we recommend installing Adobe Campaign on 2 machines, with the following specification:
2 systems, 3G + quad-core CPU,hz 4 GB RAM, RAID 1, 2 x 80GB HDD
These systems create the marketing instance Application Server, which directly supports your Campaign Console users and
executes the campaign workflows. This functionality is deployed on two identical servers for high-availability, sharing a Network-Attached
Storage (NAS) file system to enable failover.
Reverse proxies in a DMZ will load balance traffic to the servers. There is no requirement to install the Adobe Campaign software stack
on proxy machines; any reverse proxy software or network equipment can be used .[1]
Subscription opt in/out and preference center features may be provided by Campaign or by your own web site. If you choose to
implement this functionality on your web site, you must ensure that preference and subscription information is propagated to the
Campaign marketing database. This is normally done by creating extraction files which are uploaded by Campaign workflows
automatically.
Disk space consumption on Application Servers depends on the retention period of files exchanged with third party service providers (for
example, print vendors for Direct Mail), and on the size and retention of imported flat files, like subscription or preference updates from
your web site, or extracts from your own CRM or marketing systems.
Database
Hardware recommendations for the database server are as follows:
3G + 4-core CPU,hz 16 GB RAM, RAID 1, 2 x 64 GB SSD or RAID 10, 4 x 64 GB SSD
The memory estimate assumes full caching of approximately 500,000 recipients for a large campaign launch, plus RDBMS buffer space
for executing workflows, importing tracking data and other concurrent activities.
We estimate that the disk space required on the database to store all Adobe Campaign technical data (campaigns, tracking, working
tables, etc.) is based on a retention period of If you choose to retain tracking data for 6 months, database size7 GB 3 months.
increases to about 9.2 GB, and 12-month retention increases database size to 11 GB. Recipient data will consume for thisabout 1 GB
environment.
Important: This estimate does not include any additional customer data. If you are planning to replicate additional columns or tables of
customer data into the Adobe Campaign database, then you need to estimate the additional disk space requirement for this. Uploaded
segments/lists will also require additional storage, depending on their size, frequency and retention period.
Please also take into consideration that due to the volume of information processed daily, the IOPS of the database server are critical.
For instance, on a peak day, you may deploy campaigns targeting a total of 500,000 recipients. In order to execute each campaign,
Adobe Campaign will insert 500,000 records into a table containing roughy 12 million records (the delivery log table). To provide
acceptable performance during the campaign deployment, we recommend a minimum of 60,000 4KB Random Read/Write IOPS for this
scenario.
Scenario 2: Large-Size Deployment
Copyright © 2016, Adobe
Adobe Campaign Page 4
Volume Estimates
Channel Comments
Active recipients 10 million
Email 42 million/month
Direct Mail 10 million/month
Mobile SMS 1,000,000/month
Peak daily email volume 5,000,000
At larger volumes, the web traffic to on-premise Campaign systems can be significant, especially if inbound Interactions are used. To
ensure good performance for the marketing instance application servers, web application traffic is routed to separate Campaign web
servers, so bursts of web traffic after large campaign launches do not impact Campaign workflows and Console users
The deployment for Adobe Campaign changes to add two web servers for Campaign web applications. Hardware specifications are also
increased for the app servers and database to support the larger workload.
Application Servers
In this scenario, we recommend installing Adobe Campaign on 4 machines, 2 application servers and 2 web servers, with the following
specification:
4 systems, 3G + quad-core CPU,hz 8 GB RAM, RAID 1, 2 x 80GB HDD
The Application Servers directly support Campaign Console users and the execution of campaign workflows. This functionality is
deployed on two identical servers for high-availability, sharing a Network-Attached Storage (NAS) file system to enable failover.
The Web Servers host Campaign web applications supporting the 10-million active recipients in the system.
See Scenario 1 for additional comments on proxies, preference centers/subscription handling, and disk space usage.
Database
Hardware recommendations for the database server are as follows:
3G + 8-core CPU,hz 64 GB RAM RAID 1, 2 x 320 GB SSD or RAID 10, 4 x 320 GB SSD
The memory estimate assumes full caching of approximately 5,000,000 recipients for a large campaign launch, plus RDBMS buffer
space for executing workflows, importing tracking data and other concurrent activities.
We estimate that the disk space required on the database to store all Adobe Campaign technical data (campaigns, tracking, working
tables, etc.) is based on a retention period of If you choose to retain tracking data for 6 months, database size140 GB 3 months.
increases to about 242 GB, and 12-month retention increases database size to 446 GB. Recipient data will consume for thisabout 7 GB
environment.
Scenario 3: Enterprise Deployment with Message Center
Copyright © 2016, Adobe
Adobe Campaign Page 5
Volume Estimates
Channel Comments
Active recipients 25 million
Email 108 million/month
Direct Mail 25 million/month
Mobile SMS 2.5 million SMS/month
Message Center 2.5 million transactional messages/month
Peak daily email volume 2,500,000
The deployment supporting 25 million recipients is essentially the same as shown in Scenario 2: Campaign web application traffic is
routed to Campaign web servers, so the bursts of web traffic after large campaign launches does not impact Campaign workflows and
Console users.
This deployment also includes Message Center calls, driven from your own web sites and applications.
Application Servers
In this scenario, we recommend installing Adobe Campaign on 4 machines, as follows:
Application Servers
2 systems, 3G + quad-core CPU,hz 8 GB RAM, RAID 1, 2 x 160GB HDD
Web Servers
2 systems, 3G + quad-core CPU,hz 16 GB RAM, RAID 1, 2 x 80GB HDD
The Application Servers directly support Campaign Console users and the execution of campaign workflows. This functionality is
deployed on two identical servers for high-availability, sharing a Network-Attached Storage (NAS) file system to enable failover.
The Web Servers host Campaign web applications supporting the 10-million active recipients in the system.
See Scenario 1 for additional comments on proxies, preference centers/subscription handling, and disk space usage.
Database
Copyright © 2016, Adobe
Adobe Campaign Page 6
Hardware recommendations for the database server are as follows:
3G + 8-core CPU,hz 96 GB RAM, RAID 1, 2 x 768 GB SSD or RAID 10, 4 x 768 GB SSD
The memory estimate assumes full caching of approximately 12,500,000 recipients for a large campaign launch, plus RDBMS buffer
space for executing workflows, importing tracking data and other concurrent activities.
We estimate that the disk space required on the database to store all Adobe Campaign technical data (campaigns, tracking, working
tables, etc.) is based on a retention period of If you choose to retain tracking data for 6 months, database size387 GB 3 months.
increases to about 679 GB, and 12-month retention increases database size to 1262 GB. Recipient data will consume forabout 18 GB
this environment.
Guidelines for Changing Assumptions
The assumptions we made for these scenarios all had a significant impact on the hardware recommendations and deployment
architecture. This section discusses guidelines around different assumptions. Please contact the Adobe Campaign Consulting team
for specific recommendations to meet your requirements.
Number of Recipients
Active recipients require both storage space and database buffer space, so more recipients generally require more memory and CPU
capacity on the database server. Storage increases are relatively small for the recipients themselves, but may be significant for the event
tracking data kept for email campaigns.
Email Campaign Size
The number of emails in a given campaign does not primarily impact the sizing of on-premise servers (in your data center or hosted
service provider); they impact the size and number of MTAs used by Adobe Cloud Messaging to send the campaigns. However, the total
number of emails sent by the system (campaign size times campaign frequency), and the response rate of campaign recipients,
determines the amount of tracking data that is collected from ACM and stored in the customer marketing database. When you have
highly-engaged recipients who open your messages and click on links, you will receive more tracking data for analytics.
The number of emails and amount of personalization determine how quickly a campaign can be sent out. More emails and heavier
personalization, and/or increases in email size due to attachments or images, will slow down campaign delivery rate.
Email Campaign Frequency
The frequency of campaign launches has an impact on database server CPU requirements. Combined with direct mailing, inbound
Interactions, and other workflows, segmentation operations for email campaigns put a significant load on the database server.
Direct Mail Frequency
The frequency of direct mailings can impact database server CPU requirements. Combined with campaign launches and other
workflows, segmentation operations for direct mailings put a significant load on the database server.
SMS Message Volume
Like email campaign size, SMS message volume does not place large loads on Campaign servers located on-premise; the load is mostly
on Adobe Cloud Messaging servers on the cloud. Segmentation for SMS campaigns, like email and direct mail, can place a significant
load on the marketing database. Therefore the frequency of SMS campaign launches and complexity of segmentation are more relevant
than the volume of SMS messages.
Database Schema Complexity
The amount of data for each active recipient requires both storage space and database buffer space, so more recipients generally
require more memory and CPU on the database server. Complex schemas also require more tables to be joined for segmentation, so
segmentation operations can run much slower, and require more database CPU and memory when data is spread across multiple
tables.
We estimate database server memory by ensuring that the database buffer pool can be large enough to contain all recipient data, plus
Copyright © 2016, Adobe
Adobe Campaign Page 7
temporary tables for running workflows, plus a margin for other database operations.
Outbound Interaction Usage
Outbound Interaction rules and offers are evaluated in workflows on the marketing instance application server, requiring app server CPU
and memory. Complex interactions rules, or large numbers of offers can require very large amounts of CPU.
Inbound Interactions and/or SOAP API Usage
Inbound Interaction rules and offers are evaluated in the marketing database, requiring significant database server resources, especially
CPU. Heavy usage of inbound Interactions and/or SOAP APIs require separate web servers to separate the work load from running
Campaign workflows.
Tracking Data Retention Period
Increasing the retention of tracking data beyond 90 days requires more database storage, and can slow down the system because
inserting new tracking data goes into very large tables. Tracking data is generally not useful for campaign segmentation after 90 days, so
we recommend the shorter retention period.
Tracking data should be moved into Adobe Analytics or another analytics system if you require long-term analysis of recipient marketing
experience.
Virtualization
All Campaign servers are good candidates for virtualization. Several issues must be addressed to ensure appropriate availability and
performance.
Fail-Over Configuration
Clustered servers, for example, redundant application servers under a load-balanced proxy, must be deployed on separate hardware to
ensure that both VMs do not go down in the event of hardware failure.
I/O Configuration
Any recommended RAID configuration must be maintained for database security, to ensure loss of a storage device does not cause loss
of data.
I/O Performance
The recommended IOPS rating for database storage must be respected. Cloud services like Amazon EC2 may not provide the required
performance, and must be carefully evaluated. For example, Amazon EC2 provisioned SSD volumes are currently rated at 20,000 IOPS
each (see ), so a 4-volume RAID configuration would behttp://docs.aws.amazon.com/AWSEC2/latest/UserGuide/EBSVolumeTypes.html
rated at 80,000 IOPS, which may not be sufficient.
We recommend performance testing for any virtualized deployment of Adobe Campaign before putting the system into production.

More Related Content

Viewers also liked

O falcão quiriquiri na falcoaria moderna traduzido por kátia boroni
O falcão quiriquiri na falcoaria moderna traduzido por kátia boroniO falcão quiriquiri na falcoaria moderna traduzido por kátia boroni
O falcão quiriquiri na falcoaria moderna traduzido por kátia boroniKátia Boroni
 
พอทของนัท
พอทของนัทพอทของนัท
พอทของนัทdaungkamon77
 
Costo de fabricacion de la mermelada aguaymanto
Costo de fabricacion de la mermelada aguaymantoCosto de fabricacion de la mermelada aguaymanto
Costo de fabricacion de la mermelada aguaymantoKevin Arthur Veliz
 
Ge Grupo de evangelismo - Pr. Mário silveira e Pr. Anderson menger
Ge   Grupo de evangelismo - Pr. Mário silveira e Pr. Anderson mengerGe   Grupo de evangelismo - Pr. Mário silveira e Pr. Anderson menger
Ge Grupo de evangelismo - Pr. Mário silveira e Pr. Anderson mengerIgreja Seara Porto Alegre
 
Legislación laboral
Legislación laboralLegislación laboral
Legislación laboralValen20
 
Análisis Comparativo de las Organizaciones
Análisis Comparativo de las OrganizacionesAnálisis Comparativo de las Organizaciones
Análisis Comparativo de las OrganizacionesCarlos Garaycoa Díaz
 
Los manuales administrativos
Los manuales administrativosLos manuales administrativos
Los manuales administrativosNombre Apellidos
 
Organizaciones tradicionales y organizaciones actuales
Organizaciones tradicionales y organizaciones actualesOrganizaciones tradicionales y organizaciones actuales
Organizaciones tradicionales y organizaciones actualesAndrêa Gomez
 
Practica rango de celdas y graficos
Practica rango de celdas y graficos Practica rango de celdas y graficos
Practica rango de celdas y graficos maisabelparra
 
Tema 13 Transformaciones económicas y cambios sociales en el siglo XIX
Tema 13  Transformaciones económicas y cambios sociales en el siglo XIXTema 13  Transformaciones económicas y cambios sociales en el siglo XIX
Tema 13 Transformaciones económicas y cambios sociales en el siglo XIXSergio Garcia
 

Viewers also liked (17)

O falcão quiriquiri na falcoaria moderna traduzido por kátia boroni
O falcão quiriquiri na falcoaria moderna traduzido por kátia boroniO falcão quiriquiri na falcoaria moderna traduzido por kátia boroni
O falcão quiriquiri na falcoaria moderna traduzido por kátia boroni
 
TLE - Horticulture Cattle Raising
TLE - Horticulture Cattle RaisingTLE - Horticulture Cattle Raising
TLE - Horticulture Cattle Raising
 
Director research
Director researchDirector research
Director research
 
Escuela verde v3
Escuela verde v3Escuela verde v3
Escuela verde v3
 
พอทของนัท
พอทของนัทพอทของนัท
พอทของนัท
 
Visão seara - Igreja Seara Porto Alegre
Visão seara -  Igreja Seara Porto AlegreVisão seara -  Igreja Seara Porto Alegre
Visão seara - Igreja Seara Porto Alegre
 
Ensayo
EnsayoEnsayo
Ensayo
 
Costo de fabricacion de la mermelada aguaymanto
Costo de fabricacion de la mermelada aguaymantoCosto de fabricacion de la mermelada aguaymanto
Costo de fabricacion de la mermelada aguaymanto
 
Starting Your Journey In Game Localization
Starting Your Journey In Game LocalizationStarting Your Journey In Game Localization
Starting Your Journey In Game Localization
 
Ge Grupo de evangelismo - Pr. Mário silveira e Pr. Anderson menger
Ge   Grupo de evangelismo - Pr. Mário silveira e Pr. Anderson mengerGe   Grupo de evangelismo - Pr. Mário silveira e Pr. Anderson menger
Ge Grupo de evangelismo - Pr. Mário silveira e Pr. Anderson menger
 
Legislación laboral
Legislación laboralLegislación laboral
Legislación laboral
 
Análisis Comparativo de las Organizaciones
Análisis Comparativo de las OrganizacionesAnálisis Comparativo de las Organizaciones
Análisis Comparativo de las Organizaciones
 
Los manuales administrativos
Los manuales administrativosLos manuales administrativos
Los manuales administrativos
 
Organizaciones tradicionales y organizaciones actuales
Organizaciones tradicionales y organizaciones actualesOrganizaciones tradicionales y organizaciones actuales
Organizaciones tradicionales y organizaciones actuales
 
Practica rango de celdas y graficos
Practica rango de celdas y graficos Practica rango de celdas y graficos
Practica rango de celdas y graficos
 
Tema 13 Transformaciones económicas y cambios sociales en el siglo XIX
Tema 13  Transformaciones económicas y cambios sociales en el siglo XIXTema 13  Transformaciones económicas y cambios sociales en el siglo XIX
Tema 13 Transformaciones económicas y cambios sociales en el siglo XIX
 
Conceptos generales
Conceptos generalesConceptos generales
Conceptos generales
 

Similar to AdobeCampaign_HardwareSizingGuide

78425589-Cloud-Optimization-Whitepaper.pdf
78425589-Cloud-Optimization-Whitepaper.pdf78425589-Cloud-Optimization-Whitepaper.pdf
78425589-Cloud-Optimization-Whitepaper.pdfJay Kulkarni
 
SharePoint 2013 supported DB's
SharePoint 2013 supported DB'sSharePoint 2013 supported DB's
SharePoint 2013 supported DB'sak-allaire
 
An Introduction to Microsoft Flow
An Introduction to Microsoft FlowAn Introduction to Microsoft Flow
An Introduction to Microsoft FlowRobert Crane
 
Getting Started with ML in AdTech - AWS Online Tech Talks
Getting Started with ML in AdTech - AWS Online Tech TalksGetting Started with ML in AdTech - AWS Online Tech Talks
Getting Started with ML in AdTech - AWS Online Tech TalksAmazon Web Services
 
How Oath (a Verizon Company) Built a Multi-Region GDPR Application with Amazo...
How Oath (a Verizon Company) Built a Multi-Region GDPR Application with Amazo...How Oath (a Verizon Company) Built a Multi-Region GDPR Application with Amazo...
How Oath (a Verizon Company) Built a Multi-Region GDPR Application with Amazo...Amazon Web Services
 
Introduction to requirement of microservices
Introduction to requirement of microservicesIntroduction to requirement of microservices
Introduction to requirement of microservicesAvik Das
 
Jamcracker Cloud Management Platform Updates: Devops Framework, Migration Pla...
Jamcracker Cloud Management Platform Updates: Devops Framework, Migration Pla...Jamcracker Cloud Management Platform Updates: Devops Framework, Migration Pla...
Jamcracker Cloud Management Platform Updates: Devops Framework, Migration Pla...Jamcracker Inc
 
Hadoop Demo eConvergence
Hadoop Demo eConvergenceHadoop Demo eConvergence
Hadoop Demo eConvergencekvnnrao
 
AWS Enterprise Summit Netherlands - Cost Optimisation at Scale
AWS Enterprise Summit Netherlands - Cost Optimisation at ScaleAWS Enterprise Summit Netherlands - Cost Optimisation at Scale
AWS Enterprise Summit Netherlands - Cost Optimisation at ScaleAmazon Web Services
 
Build end-to-end solutions with BlueMix, Avi Vizel & Ziv Dai, IBM
Build end-to-end solutions with BlueMix, Avi Vizel & Ziv Dai, IBMBuild end-to-end solutions with BlueMix, Avi Vizel & Ziv Dai, IBM
Build end-to-end solutions with BlueMix, Avi Vizel & Ziv Dai, IBMCodemotion Tel Aviv
 
Build a Multi-Region Serverless Application for Resilience & High Availabilit...
Build a Multi-Region Serverless Application for Resilience & High Availabilit...Build a Multi-Region Serverless Application for Resilience & High Availabilit...
Build a Multi-Region Serverless Application for Resilience & High Availabilit...Amazon Web Services
 
Serverless Architectural Patterns and Best Practices (ARC305-R2) - AWS re:Inv...
Serverless Architectural Patterns and Best Practices (ARC305-R2) - AWS re:Inv...Serverless Architectural Patterns and Best Practices (ARC305-R2) - AWS re:Inv...
Serverless Architectural Patterns and Best Practices (ARC305-R2) - AWS re:Inv...Amazon Web Services
 
Gp Installation Presentation
Gp Installation PresentationGp Installation Presentation
Gp Installation Presentationguest2fc298
 
Gp Installation Presentation
Gp Installation PresentationGp Installation Presentation
Gp Installation Presentationddauphin
 
Cloud Economics; How to Quantify the Benefits of Moving to the Cloud - Transf...
Cloud Economics; How to Quantify the Benefits of Moving to the Cloud - Transf...Cloud Economics; How to Quantify the Benefits of Moving to the Cloud - Transf...
Cloud Economics; How to Quantify the Benefits of Moving to the Cloud - Transf...Amazon Web Services
 
Modernize Your Desktop and Application Delivery with AWS - AWS Online Tech Talks
Modernize Your Desktop and Application Delivery with AWS - AWS Online Tech TalksModernize Your Desktop and Application Delivery with AWS - AWS Online Tech Talks
Modernize Your Desktop and Application Delivery with AWS - AWS Online Tech TalksAmazon Web Services
 

Similar to AdobeCampaign_HardwareSizingGuide (20)

78425589-Cloud-Optimization-Whitepaper.pdf
78425589-Cloud-Optimization-Whitepaper.pdf78425589-Cloud-Optimization-Whitepaper.pdf
78425589-Cloud-Optimization-Whitepaper.pdf
 
SAPonAzureCaseStudyMay2020.pptx
SAPonAzureCaseStudyMay2020.pptxSAPonAzureCaseStudyMay2020.pptx
SAPonAzureCaseStudyMay2020.pptx
 
A Guide to the various Types of Cloud ERP in this Digital Era.pptx
A Guide to the various Types of Cloud ERP in this Digital Era.pptxA Guide to the various Types of Cloud ERP in this Digital Era.pptx
A Guide to the various Types of Cloud ERP in this Digital Era.pptx
 
SharePoint 2013 supported DB's
SharePoint 2013 supported DB'sSharePoint 2013 supported DB's
SharePoint 2013 supported DB's
 
An Introduction to Microsoft Flow
An Introduction to Microsoft FlowAn Introduction to Microsoft Flow
An Introduction to Microsoft Flow
 
Getting Started with ML in AdTech - AWS Online Tech Talks
Getting Started with ML in AdTech - AWS Online Tech TalksGetting Started with ML in AdTech - AWS Online Tech Talks
Getting Started with ML in AdTech - AWS Online Tech Talks
 
How Oath (a Verizon Company) Built a Multi-Region GDPR Application with Amazo...
How Oath (a Verizon Company) Built a Multi-Region GDPR Application with Amazo...How Oath (a Verizon Company) Built a Multi-Region GDPR Application with Amazo...
How Oath (a Verizon Company) Built a Multi-Region GDPR Application with Amazo...
 
Mdb dn 2016_12_single_view
Mdb dn 2016_12_single_viewMdb dn 2016_12_single_view
Mdb dn 2016_12_single_view
 
Introduction to requirement of microservices
Introduction to requirement of microservicesIntroduction to requirement of microservices
Introduction to requirement of microservices
 
02 power bi in a day champ
02 power bi in a day champ02 power bi in a day champ
02 power bi in a day champ
 
Jamcracker Cloud Management Platform Updates: Devops Framework, Migration Pla...
Jamcracker Cloud Management Platform Updates: Devops Framework, Migration Pla...Jamcracker Cloud Management Platform Updates: Devops Framework, Migration Pla...
Jamcracker Cloud Management Platform Updates: Devops Framework, Migration Pla...
 
Hadoop Demo eConvergence
Hadoop Demo eConvergenceHadoop Demo eConvergence
Hadoop Demo eConvergence
 
AWS Enterprise Summit Netherlands - Cost Optimisation at Scale
AWS Enterprise Summit Netherlands - Cost Optimisation at ScaleAWS Enterprise Summit Netherlands - Cost Optimisation at Scale
AWS Enterprise Summit Netherlands - Cost Optimisation at Scale
 
Build end-to-end solutions with BlueMix, Avi Vizel & Ziv Dai, IBM
Build end-to-end solutions with BlueMix, Avi Vizel & Ziv Dai, IBMBuild end-to-end solutions with BlueMix, Avi Vizel & Ziv Dai, IBM
Build end-to-end solutions with BlueMix, Avi Vizel & Ziv Dai, IBM
 
Build a Multi-Region Serverless Application for Resilience & High Availabilit...
Build a Multi-Region Serverless Application for Resilience & High Availabilit...Build a Multi-Region Serverless Application for Resilience & High Availabilit...
Build a Multi-Region Serverless Application for Resilience & High Availabilit...
 
Serverless Architectural Patterns and Best Practices (ARC305-R2) - AWS re:Inv...
Serverless Architectural Patterns and Best Practices (ARC305-R2) - AWS re:Inv...Serverless Architectural Patterns and Best Practices (ARC305-R2) - AWS re:Inv...
Serverless Architectural Patterns and Best Practices (ARC305-R2) - AWS re:Inv...
 
Gp Installation Presentation
Gp Installation PresentationGp Installation Presentation
Gp Installation Presentation
 
Gp Installation Presentation
Gp Installation PresentationGp Installation Presentation
Gp Installation Presentation
 
Cloud Economics; How to Quantify the Benefits of Moving to the Cloud - Transf...
Cloud Economics; How to Quantify the Benefits of Moving to the Cloud - Transf...Cloud Economics; How to Quantify the Benefits of Moving to the Cloud - Transf...
Cloud Economics; How to Quantify the Benefits of Moving to the Cloud - Transf...
 
Modernize Your Desktop and Application Delivery with AWS - AWS Online Tech Talks
Modernize Your Desktop and Application Delivery with AWS - AWS Online Tech TalksModernize Your Desktop and Application Delivery with AWS - AWS Online Tech Talks
Modernize Your Desktop and Application Delivery with AWS - AWS Online Tech Talks
 

AdobeCampaign_HardwareSizingGuide

  • 1. Copyright © 2016, Adobe Adobe Campaign Page 1 1. 2. 3. Hardware Sizing Guide Overview This document provides general guidelines for Adobe Campaign Version 6 deployment at your on-premise data center or virtualized cloud environment. This type of deployment, referred to as “hybrid” or “mid-sourcing”, places the Campaign marketing instance and marketing database under your operational control, while using Adobe Cloud Messaging services to send emails, SMS and/or SMPP messages, and collect email open, bounce and click tracking data. The marketing instance is the portion of the Adobe Campaign architecture that drives all marketing activity, and stores all recipient data and analytic data returned by campaigns. The marketing instance is created by a set of on-premise servers running Adobe Campaign services, and a relational database. The recommendations in this document do not apply if you are using a fully-hosted Adobe Campaign instance (deployed in Adobe Cloud Services), nor do they apply to Adobe Campaign Standard edition. Scenarios Deployment diagrams and hardware sizing recommendations are provided for three representative scenarios: Moderate Size – 1 million active recipients in the system, Large Size – 10 million active recipients in the system, and Enterprise – 25 million active recipients, with additional transactional messaging. Assumptions We also assume the following types of usage for all three scenarios: Large email campaigns are sent twice each week, to approximately 50% of your active recipients Direct mailings are generated once per month for each recipient in the system SMS messages are sent to approximately 10% of your active recipients each month The database schema defining each recipient has been extended with one additional table, containing about two hundred bytes of data for each recipient Adobe Campaign Interactions are used to add offers to outgoing email Email tracking data is retained in the Campaign system for 90 days Compatibility Matrix See https://docs.campaign.adobe.com/doc/AC6.1/en/technicalResources/matrix.html General Guidelines Campaign is a database-centric application, and database server performance is critical. Running workflows, segmentation, tracking data uploads, inbound Interactions, analytics and other activities all generate database activity. In general, the size and frequency of these operations determine the size of your database server(s). The application servers in your marketing instance require enough CPU and memory to run workflows and respond to SOAP API calls, including requests from Campaign Console users. CPU requirements can be significant for workflows that use outbound Interactions with complex offer rules, workflows that execute custom Javascript, and web applications with high traffic levels. Document Purpose General guidelines for estimating hardware requirements for on-premise Campaign deployment. This document is also attached as a PDF, which may be shared with customers.
  • 2. Copyright © 2016, Adobe Adobe Campaign Page 2 Campaign web applications may also be deployed on the marketing instance app servers, or on separate web server systems. Since web application workloads conflict with critical workflows and Campaign Console users, web applications and inbound Interactions may be deployed to separate servers, to ensure that core Campaign functionality runs reliably with good performance. Reverse Proxies The Campaign architecture enforces high security by using SSL over HTTP (HTTPS) to communicate between your marketing instance and Adobe Cloud Messaging. Security, reliability and availability are enforced by using reverse proxies in a “demilitarized zone” (DMZ) subnet to isolate and secure the marketing instance servers and database. When deploying proxies on your own systems, the following hardware recommendation is sufficient for all scenarios described in this document: 2G + dual-core CPU,hz 4 GB RAM, RAID 1, 2 x 80GB HDD Notes: Adobe Campaign includes a reverse proxy which may be embedded into IIS or Apache; the hardware recommendation above is applicable only if you decide to go with the Adobe Campaign reverse proxy. The Application Server proxy is set up in an active/passive configuration, with HTTPS terminated at the proxy. Adobe will provide you with the exclusive list of URLs that should be relayed to the Adobe Campaign server in your deployment environment. Scenario 1: Moderate-Size Deployment Volume Estimates Channel Comments Active recipients 1 million Email 4.2 million/month Direct Mail 1 million/month Mobile SMS 100,000/month Peak daily email volume 500,000 For these volumes, a pair of Adobe Campaign application server systems provide all functionality for Adobe Campaign Client users, Campaign web applications and workflow execution. For 1 million active recipients and this email volume, application server workloads are not CPU or I/O intensive; most of the stress is on the database. Your web servers are shown in the secure zone, showing how recipient clicks on links in email come back to your system. In fact, these links get redirected through Adobe Cloud Messaging, so they can be tracked. Application Servers
  • 3. Copyright © 2016, Adobe Adobe Campaign Page 3 In this scenario, we recommend installing Adobe Campaign on 2 machines, with the following specification: 2 systems, 3G + quad-core CPU,hz 4 GB RAM, RAID 1, 2 x 80GB HDD These systems create the marketing instance Application Server, which directly supports your Campaign Console users and executes the campaign workflows. This functionality is deployed on two identical servers for high-availability, sharing a Network-Attached Storage (NAS) file system to enable failover. Reverse proxies in a DMZ will load balance traffic to the servers. There is no requirement to install the Adobe Campaign software stack on proxy machines; any reverse proxy software or network equipment can be used .[1] Subscription opt in/out and preference center features may be provided by Campaign or by your own web site. If you choose to implement this functionality on your web site, you must ensure that preference and subscription information is propagated to the Campaign marketing database. This is normally done by creating extraction files which are uploaded by Campaign workflows automatically. Disk space consumption on Application Servers depends on the retention period of files exchanged with third party service providers (for example, print vendors for Direct Mail), and on the size and retention of imported flat files, like subscription or preference updates from your web site, or extracts from your own CRM or marketing systems. Database Hardware recommendations for the database server are as follows: 3G + 4-core CPU,hz 16 GB RAM, RAID 1, 2 x 64 GB SSD or RAID 10, 4 x 64 GB SSD The memory estimate assumes full caching of approximately 500,000 recipients for a large campaign launch, plus RDBMS buffer space for executing workflows, importing tracking data and other concurrent activities. We estimate that the disk space required on the database to store all Adobe Campaign technical data (campaigns, tracking, working tables, etc.) is based on a retention period of If you choose to retain tracking data for 6 months, database size7 GB 3 months. increases to about 9.2 GB, and 12-month retention increases database size to 11 GB. Recipient data will consume for thisabout 1 GB environment. Important: This estimate does not include any additional customer data. If you are planning to replicate additional columns or tables of customer data into the Adobe Campaign database, then you need to estimate the additional disk space requirement for this. Uploaded segments/lists will also require additional storage, depending on their size, frequency and retention period. Please also take into consideration that due to the volume of information processed daily, the IOPS of the database server are critical. For instance, on a peak day, you may deploy campaigns targeting a total of 500,000 recipients. In order to execute each campaign, Adobe Campaign will insert 500,000 records into a table containing roughy 12 million records (the delivery log table). To provide acceptable performance during the campaign deployment, we recommend a minimum of 60,000 4KB Random Read/Write IOPS for this scenario. Scenario 2: Large-Size Deployment
  • 4. Copyright © 2016, Adobe Adobe Campaign Page 4 Volume Estimates Channel Comments Active recipients 10 million Email 42 million/month Direct Mail 10 million/month Mobile SMS 1,000,000/month Peak daily email volume 5,000,000 At larger volumes, the web traffic to on-premise Campaign systems can be significant, especially if inbound Interactions are used. To ensure good performance for the marketing instance application servers, web application traffic is routed to separate Campaign web servers, so bursts of web traffic after large campaign launches do not impact Campaign workflows and Console users The deployment for Adobe Campaign changes to add two web servers for Campaign web applications. Hardware specifications are also increased for the app servers and database to support the larger workload. Application Servers In this scenario, we recommend installing Adobe Campaign on 4 machines, 2 application servers and 2 web servers, with the following specification: 4 systems, 3G + quad-core CPU,hz 8 GB RAM, RAID 1, 2 x 80GB HDD The Application Servers directly support Campaign Console users and the execution of campaign workflows. This functionality is deployed on two identical servers for high-availability, sharing a Network-Attached Storage (NAS) file system to enable failover. The Web Servers host Campaign web applications supporting the 10-million active recipients in the system. See Scenario 1 for additional comments on proxies, preference centers/subscription handling, and disk space usage. Database Hardware recommendations for the database server are as follows: 3G + 8-core CPU,hz 64 GB RAM RAID 1, 2 x 320 GB SSD or RAID 10, 4 x 320 GB SSD The memory estimate assumes full caching of approximately 5,000,000 recipients for a large campaign launch, plus RDBMS buffer space for executing workflows, importing tracking data and other concurrent activities. We estimate that the disk space required on the database to store all Adobe Campaign technical data (campaigns, tracking, working tables, etc.) is based on a retention period of If you choose to retain tracking data for 6 months, database size140 GB 3 months. increases to about 242 GB, and 12-month retention increases database size to 446 GB. Recipient data will consume for thisabout 7 GB environment. Scenario 3: Enterprise Deployment with Message Center
  • 5. Copyright © 2016, Adobe Adobe Campaign Page 5 Volume Estimates Channel Comments Active recipients 25 million Email 108 million/month Direct Mail 25 million/month Mobile SMS 2.5 million SMS/month Message Center 2.5 million transactional messages/month Peak daily email volume 2,500,000 The deployment supporting 25 million recipients is essentially the same as shown in Scenario 2: Campaign web application traffic is routed to Campaign web servers, so the bursts of web traffic after large campaign launches does not impact Campaign workflows and Console users. This deployment also includes Message Center calls, driven from your own web sites and applications. Application Servers In this scenario, we recommend installing Adobe Campaign on 4 machines, as follows: Application Servers 2 systems, 3G + quad-core CPU,hz 8 GB RAM, RAID 1, 2 x 160GB HDD Web Servers 2 systems, 3G + quad-core CPU,hz 16 GB RAM, RAID 1, 2 x 80GB HDD The Application Servers directly support Campaign Console users and the execution of campaign workflows. This functionality is deployed on two identical servers for high-availability, sharing a Network-Attached Storage (NAS) file system to enable failover. The Web Servers host Campaign web applications supporting the 10-million active recipients in the system. See Scenario 1 for additional comments on proxies, preference centers/subscription handling, and disk space usage. Database
  • 6. Copyright © 2016, Adobe Adobe Campaign Page 6 Hardware recommendations for the database server are as follows: 3G + 8-core CPU,hz 96 GB RAM, RAID 1, 2 x 768 GB SSD or RAID 10, 4 x 768 GB SSD The memory estimate assumes full caching of approximately 12,500,000 recipients for a large campaign launch, plus RDBMS buffer space for executing workflows, importing tracking data and other concurrent activities. We estimate that the disk space required on the database to store all Adobe Campaign technical data (campaigns, tracking, working tables, etc.) is based on a retention period of If you choose to retain tracking data for 6 months, database size387 GB 3 months. increases to about 679 GB, and 12-month retention increases database size to 1262 GB. Recipient data will consume forabout 18 GB this environment. Guidelines for Changing Assumptions The assumptions we made for these scenarios all had a significant impact on the hardware recommendations and deployment architecture. This section discusses guidelines around different assumptions. Please contact the Adobe Campaign Consulting team for specific recommendations to meet your requirements. Number of Recipients Active recipients require both storage space and database buffer space, so more recipients generally require more memory and CPU capacity on the database server. Storage increases are relatively small for the recipients themselves, but may be significant for the event tracking data kept for email campaigns. Email Campaign Size The number of emails in a given campaign does not primarily impact the sizing of on-premise servers (in your data center or hosted service provider); they impact the size and number of MTAs used by Adobe Cloud Messaging to send the campaigns. However, the total number of emails sent by the system (campaign size times campaign frequency), and the response rate of campaign recipients, determines the amount of tracking data that is collected from ACM and stored in the customer marketing database. When you have highly-engaged recipients who open your messages and click on links, you will receive more tracking data for analytics. The number of emails and amount of personalization determine how quickly a campaign can be sent out. More emails and heavier personalization, and/or increases in email size due to attachments or images, will slow down campaign delivery rate. Email Campaign Frequency The frequency of campaign launches has an impact on database server CPU requirements. Combined with direct mailing, inbound Interactions, and other workflows, segmentation operations for email campaigns put a significant load on the database server. Direct Mail Frequency The frequency of direct mailings can impact database server CPU requirements. Combined with campaign launches and other workflows, segmentation operations for direct mailings put a significant load on the database server. SMS Message Volume Like email campaign size, SMS message volume does not place large loads on Campaign servers located on-premise; the load is mostly on Adobe Cloud Messaging servers on the cloud. Segmentation for SMS campaigns, like email and direct mail, can place a significant load on the marketing database. Therefore the frequency of SMS campaign launches and complexity of segmentation are more relevant than the volume of SMS messages. Database Schema Complexity The amount of data for each active recipient requires both storage space and database buffer space, so more recipients generally require more memory and CPU on the database server. Complex schemas also require more tables to be joined for segmentation, so segmentation operations can run much slower, and require more database CPU and memory when data is spread across multiple tables. We estimate database server memory by ensuring that the database buffer pool can be large enough to contain all recipient data, plus
  • 7. Copyright © 2016, Adobe Adobe Campaign Page 7 temporary tables for running workflows, plus a margin for other database operations. Outbound Interaction Usage Outbound Interaction rules and offers are evaluated in workflows on the marketing instance application server, requiring app server CPU and memory. Complex interactions rules, or large numbers of offers can require very large amounts of CPU. Inbound Interactions and/or SOAP API Usage Inbound Interaction rules and offers are evaluated in the marketing database, requiring significant database server resources, especially CPU. Heavy usage of inbound Interactions and/or SOAP APIs require separate web servers to separate the work load from running Campaign workflows. Tracking Data Retention Period Increasing the retention of tracking data beyond 90 days requires more database storage, and can slow down the system because inserting new tracking data goes into very large tables. Tracking data is generally not useful for campaign segmentation after 90 days, so we recommend the shorter retention period. Tracking data should be moved into Adobe Analytics or another analytics system if you require long-term analysis of recipient marketing experience. Virtualization All Campaign servers are good candidates for virtualization. Several issues must be addressed to ensure appropriate availability and performance. Fail-Over Configuration Clustered servers, for example, redundant application servers under a load-balanced proxy, must be deployed on separate hardware to ensure that both VMs do not go down in the event of hardware failure. I/O Configuration Any recommended RAID configuration must be maintained for database security, to ensure loss of a storage device does not cause loss of data. I/O Performance The recommended IOPS rating for database storage must be respected. Cloud services like Amazon EC2 may not provide the required performance, and must be carefully evaluated. For example, Amazon EC2 provisioned SSD volumes are currently rated at 20,000 IOPS each (see ), so a 4-volume RAID configuration would behttp://docs.aws.amazon.com/AWSEC2/latest/UserGuide/EBSVolumeTypes.html rated at 80,000 IOPS, which may not be sufficient. We recommend performance testing for any virtualized deployment of Adobe Campaign before putting the system into production.