Transcript IBM Mobile Enterprise Exec Overview

IBM Cloud: Think it. Build it. Tap into it.
Jueves de Softlayer
Carlos Aguado Fidalgo
IBM Cloud Technical Sales
© 2015 IBM Corporation
Agenda
1. Introducción al Cloud – Modelo IaaS
2. IBM Softlayer
1. Computación
2. Almacenamiento
3. Redes
4. Seguridad
5. Soluciones (Disaster Recovery, Backup, etc)
3. Demo práctica del Portal
© 2015 IBM Corporation
¿Qué puede aportar Cloud?
Rápido
aprovisionamiento
1
2
3
4
Pago
por uso
Accesibilidad
Elasticidad
Acelerar el despliegue de nuevos productos y servicios
Transformación de un modelo CAPEX a OPEX
Facilitar la expansión internacional
Manejar completamente tu infraestructura
© 2015 IBM Corporation
capacity
Beneficios del cloud – IT tradicional vs Cloud
undercapacity
overcapacity
flexible capacity
reduction of
Initial
investment
undercapacity
LOAD
forecasted load
actual load
INFRASTRUCTURE PLANNING
traditional IT infrastructure
dynamic cloud infrastructure
time
© 2015 IBM Corporation
Virtual Servers
Virtual Server Basics
 Shared (public node) or private (private node) environment
 Run on Citrix XenServer hypervisor
 Hypervisor managed by SoftLayer, client sees OS level
and above
 Provisioned with Local or SAN-based storage
 Monthly or hourly options available
 Depolyed in 30 minutes or less
Local vs. SAN-based Storage
 Local Storage
− Best performance
− Disks in RAID10
− Ideal for low-transactional operations
− No high availability options
 SAN-based Storage
− Default storage option for Virtual Servers
− High availability
− Ideal for light, front-end operations
− Ability to assign larger volumes
Private Clouds on SoftLayer
 Bare Metal Server provisioned with a customer-specified
hypervisor
 Customer manages hypervisor and all VMs provisioned on
the hypervisor
 Available with the following SoftLayer-supported hypervisors:
− Citrix XenServer
− Microsoft Hyper-V
− VMWare
 Customer is responsible for any associated fees with
licensing on hypervisors, including operating systems
for VMs.
Helpful Links
 Virtual Servers topic page on KnowledgeLayer
 Set Up a Virtual Server on KnowledgeLayer
 Set Up a Hypervisor on KnowledgeLayer
 Virtual Servers page on softlayer.com
© 2015 IBM Corporation
Bare Metal Servers
Bare Metal Basics
 Physical device
 Single-tenant environment
 Most powerful server available
 Custom-provisioned in 2-4 hours
 Managed through Customer Portal or API, controlled
directly on device through IPMI, KVM/IP, or SSH/RDP
 24x7x365 support comes standard
Chassis, Processor, Drives
 Available in all Data Centers
 Intel-powered, x86 servers
 Single, Dual, or Quad Proc models
 Chassis Sizes
− 1U = 4 drives
− 2U = 12 drives
− 3U = ~12 drives (GPU only)
− 4U = 36 drives
 Drive Types
− SATA
− SAS-SCSI
− SSD
 Redundant Power on all 2U, 3U, and 4U servers
Network Features
 Standard
− 500 GB Outbound Public Network Bandwidth
− Single-path network connections on public, private, and
management
− 100 Mbps connectivity in all Data Centers
 Upgrades
− Dual-path network connections
− Upgraded port speed from 100 Mbps to 1GB or 10 GB.
− Bandwidth pooling for two or more severs on an
account
− Unlimited Public Network bandwidth
− Provisioning with private network only connectivity
(public network ports shut down at time of initial
configuration)
Helpful Links
 Bare Metal Server page on softlayer.com
 Bare Metal Server topic page on KnowledgeLayer
 Set up a Bare Metal Server on KnowledgeLayer
© 2015 IBM Corporation
Fixed Configuration
Fixed Configuration
Supported Operating Systems
Operating System
Hourly Availability
Monthly Availability
CentOS 5 (32-bit)
X
X
CentOS 5 (64-bit)
CentOS 6.0 (32-bit)
CentOS 6.0 (64-bit)
FreeBSD 10.x (32-bit)
Free BSD 10.x (64-bit)
FreeBSD 9 Latest (32-bit)
FreeBSD 9 Latest (64-bit)
Ubuntu 14.04
VMWare ESXi 5.5
Red Hat Enterprise Linux 5 (32-bit) with per-processor licensing
Red Hat Enterprise Linux 5 (64-bit) with per-processor licensing
Red Hat Enterprise Linx 6 (32-bit) with per-processor licensing
Red Hat Enterprise Linux 6 (64-bit) with per-processor licensing
Windows Server 2008 Enterprise Edition SP2 (32-bit)
Windows Server 2008 Enterprise Edition SP2 (64-bit)
Windows Server 2008 R2 Enterprise Edition (64-bit)
Windows Server 2008 R2 Stanrdard Edition (64-bit)
Windows Server 2008 R2 Web Edition (64-bit)
Windows Server 2012 R2 Standard Edition (64-bit)
WIndows Server 2012 Standard Edition
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
Use Case
 Batch Anaytics or Processing
 Software certify an application on a HW config
 Performance Testing
Helpful Links
 Bare Metal Server page on softlayer.com
(select Hourly tab)
 Bare Metal Server topic page on
KnowledgeLayer
 Set up a Bare Metal Server on KnowledgeLayer
© 2015 IBM Corporation
High Performance Computing with GPUs
NVidia K2 GRID








Maximum 2 Cards per 3u Server
Remote Visualization/ 3D Rendering
Virtualization of GPU Functionality
2 X Kepler GPUs (GK104)
3072 NVidia Cores (1536 per GPU)
8 GB GDDR5
Can be used for GRID Gaming with updated driver from NVidia
Use Case: Network delivered GPU acceleration for Gaming,
Virtual Desktop and Cloud Application
Upgrades
 Dual-path network connections up to 10Gbps
 Bandwidth pooling for two or more severs on an account
 Unlimited Public Network bandwidth
 Provisioning with private network only connectivity (public
network ports shut down at time of initial configuration)
 2nd GPU Card
 Up to 8 internal HDDs with choice of SATA, SAS or SSD
NVidia K10 Tesla
Helpful Links











Maximum 2 Cards per 3u Server
Purpose Built GPGPU
Cuda Parallel Computing
2 X Kepler GPUs (GK104)
3072 Nvidia Cores (1536 per GPU)
8 GB GDDR5
Use Cases
− General Purpose GPU – HPC
− Hundreds of applications accelerated
− Computational Chemistry
− Computational Finance
− Fluid Dynamics
− Medical Imaging
− Weather and Climate
− Machine Learning
NVidia Grid Virtualization for Enterprise on NVidia.com
NVidia GRID for Gaming: on Nvidia.com
High Performance Computing with Tesla on Nvidia.com
GPU Applications by Domain on Nvidia.com
© 2015 IBM Corporation
Self-managed virtualized environments
Available Hypervisors
 Citrix Xen
− Extensive API
− Features include XenMotion, HA, AWB,/HPM
− Licenses per server directly from Citrix
 Microsoft Hyper-V
− API Available (WMI)
− Features include Live migration, host clustering for free
− SL offers Hyper-V with Datacenter Edition
− No Per VM fee
 VMware
− Extensive API (VIX API)
− Extensive management tools
− Features include vMotion, Storage Motion, DRS, PM (based on license)
− SL licenses per RAM reserved for VMs
 Virtuozzo
− OS Virtualization, Para Virtualized
− Guest OS must be based on primary OS
− VM’s created at the OS Level
© 2015 IBM Corporation
Auto Scale
Features
 Governed by Scale Manage privilege.
 Groups may be given unique names for quick identification.
 Groups assigned to regions, then narrowed to data center
or VLAN.
 Member counts capped by max and min, with optional
desired member count to immediately scale the group
to the desired count.
 Assets used for metric collection by monitoring robots to
trigger scaling policies.
 Up-to-date group status indicates scaling activity for the
group in real time.
 Optional VLAN integration with multi-VLAN support
 Optional Load Balancer integration when LB already exists
and is identified when building the Auto Scale Group.
Scalable Triggers
 Based on time,
 CPU utilization
 Bandwidth levels coming in or out of the device.
 The CPU and bandwidth scaling methods utilize
monitoring agents to trigger policies.
 Optional cool down may be added to any policy to
minimize false triggers
How it works
 Group > Policy > Trigger format.
 Group consists of one or more policies with optional triggers
and cool down periods
 Policies indicate scaling actions and are invoked by triggers.
 Triggers are conditions that may be satisfied once, recurring
or by monitored resource..
Management Details
 Create Auto Scale Groups, Policies and Cooldowns
 Monitor the status of scaling group and activities
 Edit the scaling groups, policies and triggers
 Delete a group and cancel all associated members
 Grant or revoke Auto Scale privileges for sub-users through
“Scale Manage” privilege
Helpful Links
 Auto Scale with SL API
 Auto Scale on KnowledgeLayer
© 2015 IBM Corporation
Storage
Local / Server Disk
Software Defined Storage
 High-capacity bare metal servers with up to 36 drives per server,
options include:
− SATA: cost effective storage up to 6TB drives, typical use cases
are for file servers
− SAS: highest performing spinning disks up to 600GB per drive,
suitable for a variety of uses
− SSD: highest performing local disk option with up to 800GB per
sold state disk
 QuantaStor (OSNEXUS)
− iSCSI SAN and NAS access
− Installed on Bare Metal with local disks
− Customer controls all aspect of storage (design/build your own
storage solution)
− Single tenant environment
− Supports thin provisioning, I/O optimization, replication, encryption,
compression, HA via CEPH and GlusterFS, etc.
Endurance Storage
Performance Storage





Highly durable and resilient enterprise grade storage
iSCSI block level access (HA connectivity via MPIO)
File level access – NFS (HA connectivity via routed TCP/IP)
Volume sizes ranging from 20GB to 12TB
IOPS tiers - 0.25 IOPS/GB (low I/O), 2 IOPS/GB (general purpose), 4
IOPS/GB (high intensity - ~49K IOPS w/ 12TB LUN)
 Can stripe across LUNs to achieve larger volumes and IOPs
 Allows for concurrent access from multiple servers (e.g. clustered
databases)
 Optional (hourly, daily, weekly) snapshot and replication (across data
centers)







FTP / NAS
Object Storage
 Scalable file level storage
 Ideal for managing and sharing data between systems (e.g. archival,
backups, and less frequently accessed files)
 No replication options
 Dynamic, pay-as-you-go pricing
 Ideal for archive, backups, streaming media, web and static
content delivery
 Searchable, taggable
 Replicated across at least 3 servers within each data center
 Accessible via Customer Portal or REST API (integrated with CDN –
delivered via Edgecast)
Highly durable and resilient enterprise grade storage
iSCSI block level access (HA connectivity via MPIO)
File level access – NFS (HA connectivity via routed TCP/IP)
Volume sizes ranging from 20GB to 12TB
IOPS ranging from 100 to 6,000
Can stripe across LUNs to achieve larger volumes and IOPs
Ideal for most applications requiring predictable levels of performance
© 2015 IBM Corporation
Backups
EVault
Idera
 Backs up data at file system level (file level)
 Web console for each agent
 “Set-and-Forget” configuration (hourly, daily, weekly,
custom)
 Only accessible from SoftLayer servers
 Back-ended by enterprise grade storage managed by
SoftLayer
 Data can be encrypted and compressed in transit and
at rest
 Multi-site backup optional
 Disk to disk backup solution (block level)
 Software installed on Bare Metal server with internal disks
 Agent based; first backup is a full backup with incremental
forever after; each new backup is a recovery point (daily,
weekly, monthly, etc.)
 Accessible from non-SoftLayer servers
 Single tenant, customer-controlled environment
 Single interface for all backups and devices including selfmanaged VMs
 Data can be encrypted and compressed
 Multi-site backup optional
Storage
Bring Your Own
 Most of the SoftLayer storage offerings (e.g. QuantaStor,
NAS, Endurance Storage and Object Storage) can be
utilized as backup storage
 Use the servers and storage provided by SoftLayer to
build/bring your own backup solution, for example Tivoli
Storage Manager including TSM Node Replication,
Veeam, etc.
© 2015 IBM Corporation
Network Components at SoftLayer




Generic Network Layout
Loadbalancers
Firewalls
Gateway: Vyatta
© 2015 IBM Corporation
The expanding global cloud footprint offers you access to our
infrastructure choices through 40 data centers and network points
of presence for improved global reach and performance
 40 data centers worldwide
in 15 countries and five
continents globally –
connecting CMS locations to
SL locations
 Geographically dispersed
points of presence (PoPs)
bring connectivity to our
carrier-grade netwoark
backbone closer to the end
user
 These PoPs, combined with
over 2,000 Gbps connectivity
between our data centers
around the world, can deliver
exceptional bandwidth and
connectivity
LondonAmsterdam
Seattle
China
Tokyo
San Jose
Los Angeles
ChicagoToronto
Denver
Dallas
Frankfurt
Montreal
Paris
New York City
Washington D.C.
Atlanta
Houston
India
Miami
Hong Kong
Mexico City
Singapore
Brazil
CMS Data Center
Sydney
Melbourne
Data Center &
Network Point of
Presence
Network Point of Presence
© 2015 IBM Corporation
Triple-network architecture
 High-performance public network with transit from multiple tier-1 carriers
 Secure OOB management via VPN
 Private network for intra-application and inter-facility communications,
access to shared services
 Native IPv6 support
 Virtual racks for integrated management
 Complete suite of network services
© 2015 IBM Corporation
SoftLayer Network Overview
Network Details
© 2015 IBM Corporation
Network Planning
SoftLayer® IP backbone: Looking Glass offering
 SoftLayer Looking Glass is an interactive resource that is available to the public. Looking Glass can be used to test latency between SoftLayer data
centers, and between a router location and target address, subnet or hostname.
 Latency tests between a location and its target are conducted using a variety of methods, including ping and traceroute commands. Users can also
test network latency through a variety of test downloads based on size and data center location. Users operate Looking Glass under the premise
that only manual queries will be made.
 SoftLayer Looking Glass is available at http://lg.softlayer.com/
© 2015 IBM Corporation
Load Balancers
Local Load Balancers
NetScaler VPX
 Multitenant
 Proxy Connections – traffic passes through load balancer for
the duration of the connection
 Single Virtual IP (VIP)
 Pricing based on connections per second
 Up to Layer 4 load balancing
 SSL Offloading optional
 Ideal for fewer than 5000 connections/second
 No high availability option
 Public network traffic only







Dedicated Load Balancers with SSL







Single Tenant
Proxy Connections
Single VIP, with additional VIPs available by request
High availability optional (at additional fee)
Up to Layer 4 load balancing
2600 <15,000/sec connections; 6600 < 100000/sec
Public network traffic only
Managed in GUI
Utilizes static or portable IP addresses
High Availability Optional (at additional fee – 2x cost)
Up to Layer 7 load balancing
Database balancing
Global Load Balancing (GSLB) option requires Platinum License
Public and Private Network traffic
NetScaler MPX







Features similar to VPX
Managed in GUI
Public and Private Network traffic
High Availability Optional (at additional fee – 2x cost)
Requires approval from networking
Racked on available COD rack
For customers that want full control and connections over
100000/sec
 Different prices for different versions
Helpful Links
 Load Balancing Topic Page in KnowledgeLayer
 Global Load Balancing Options in KnowlegeLayer
 NetScaler VPX Data Sheet (Citrix)
© 2015 IBM Corporation
Firewall Quick Reference Guide
Standard Single Server Firewall (Multitenant)
Gateway: Vyatta
 Secure single server
 Managed via portal
 Control inbound rule set
 HA options
 Secures multiple VLANs in the same server room (private
network)
 GUI console access
 Control inbound/outbound
Dedicated Hardware Firewall (Single tenant)




HA option (Active/Passive)
Secures all servers in a single VLAN
Managed via portal
Control inbound rule set
Dedicated Fortigate Security Appliance







HA option
Secures all servers in a single VLAN
GUI console access
Control inbound/outbound
Application Layer scanning
Need to be ordered separately
1Gbps firewall with redundant links
Quick Comparison
Secure 1
or more
VLANs
Traffic
control
High
Availability
Portal
access
Standard
firewall
Single
server
Inbound
only
N/A

Dedicated
firewall
Single
public
VLAN
Inbound
only
Active
/Passive

Fortigate
appliance
Single public
VLAN


GUI
Console
Vyatta
Gateway



GUI
Console
Pricing
http://www.softlayer.com/firewalls
© 2015 IBM Corporation
Gateway: Vyatta
Overview:
 Customer configurable routing device
 Placed in front of customer specified VLANs
 Servers on that VLAN route through the Vyatta Gateway
Appliance as the first network hop.
 Configured to do almost anything that can be done with a Cisco
ASA or Juniper device such as routing, NAT, Firewall, Logging,
and VPN termination
Use Cases:
VPN Termination via the Public Network: Support for IPSec
VPN, SSL-VPN, and GRE can extend customer’s corporate
network.
NAT Based Configurations
− Allow access “private network only” devices.
Robust Firewall Requirements
− Stateful packet inspection, reporting, traffic monitoring can all
be accomplished
Multi-Tiered Networking
− DMZs, or the ability to separate servers on separate VLANs
Greater Control
© 2015 IBM Corporation
CDN
CDN Basics
 SoftLayer partners with Edgecast for all Content Delivery services.
 As of July 2014 Edgecast routes content through 44 geographically
diverse nodes.
 Charged on a per GB basis for bandwidth.
 Offer standard and SSL delivery options
Global Standard vs Global Premium
 Global Standard is the default for most CDN services. Delivers from
SoftLayer POPs around the globe.
 Global Premium is required for premium transit routes, including Vietnam,
Malaysia, Philippines, Korea, and other Asia-Pacific countries.
HTTP Large Object vs HTTP Small Object
 HTTP Large Object
− Videos and Objects over 300k
− Utilizes SSD for storage
− Delivers peak performance and fast start times
 HTTP Small Object
− Designed for objects under 300k
− Delivers content directly from RAM
− Ideal for HTML, CSS, Java
© 2015 IBM Corporation
Monitoring
Monitoring Basics
Additional Monitoring Options
 Clients must configure monitors and alerts on each device or account.
 Automated monitoring tickets are customer facing only.
 Automated Reboot from Monitoring: Monitoring system automatically
reboots the server when a monitoring alert is issued.
 24/7/365 NOC Monitoring: A personal notification will be sent in the
event a monitoring alert is issued.
TCP Monitoring
Nimsoft Monitoring
 Complimentary TCP includes PING monitoring only.
− Default setting:
• Public IP with 0 second delay
• Recommended that clients update default to avoid
false positives.
− Advanced TCP Options (additional fee)
• DNS: Generic NSLOOKUP on address
• DNS Custom: NSLOOKUP on specific domain
• FTP: FTP connection on Port 21
• HTTP: HTTP connection on Port 80
• HTTP Custom: HTTP connection on Port 80, with check
for given response text
• HTTPS: HTTP connection on Port 443
• IMAP: IMAP connection on address
• LDAP: LDAP connection on address
• NNTP: NTTP connection on address
• POP: POP connection on address
• SMTP: SMTP connection on address
• SSH: SSH connection on Port 22
• TCP Custom: TCP connection to specified port
• TELNET: TELNET connection on Port 23
• UDP SIP: UDP connection to specified port
 Basic Monitoring: Hardware and OS
− CPU, Disk, and Memory
− Process
− Remote System and Windows Services
 Advanced Monitoring: System Health
− DHCP
− File Checking
− LDAP
− Mounted File Systems
− NTP
− Network Traffic
− Performance
− SNMP
 Premium Monitoring: Application
− Apache
− DNS
− Email
− IIS
− MSSQL
− MYSQL
− TomCat
© 2015 IBM Corporation
SoftLayer API at a glance
SoftLayer [main] API
Basic Concepts:

Version 3.0 API provides 2,200 function calls to over 180
services

Endpoint (public or private)

Service (e.g. Account, Storage..)

Supports REST, SOAP & XML-RPC interfaces

Method or Action (e.g. getUsers())

Clients implemented in C#, Perl, PHP, Python, Ruby, VB .Net,
command line Python client

Data Types (String, boolean, int and Complex)


Properties (local, relational, count)
Is a major direct source of revenue


Result limits (rows filtering)
Accessible from mobile devices as well

Object Masks (properties filtering)
Object Storage API:
To begin using the API:

Based on OpenStack Swift

Enable the API use from Customer Portal
REST-based

Create unique secret API key
Clients: Java, Ruby, PHP, Python

Authentication based on user and API key
Supports Containers and Objects

Choose language client library
Content Delivery Network integration

Search integration
Message Queue API:

REST-based

64k max

Key/value pair message fields

Clients: C#, Java, Ruby, PHP, Python

Supports Topics, Topic Tags, Subscriptions

Persistent via Cloudant data layer
© 2015 IBM Corporation
Support
Support Levels
Self-Enabled Support
 24x7 support via phone, chat, and ticket
 Automation built-in to Customer Portal
 Level 1 Support (CST)
 API available to authorized users on each account
− Frontline support team
− Handles common client issues
− Escalates complex issues to Level 2 Support
 Level 2 Support (CSA)
− Escalation team
− Handles complex client issues
− Highly certified in a variety of systems
 Managed Hosting
− MySQL and MSSQL Database Administrators
− Systems Administrators
− Handles various tasks, including:
− Patching
− Backups
− Implementation
− Monitoring
Notifications
 Event Subscriptions
− Unplanned Incidents (UIP)
− RAID Alerts
− Configured through Customer Portal
 Attributes
−
−
−
−
−
−
−
Internal configuration
Sends notifications when created or moved
Supported by Level 1, Level 2, and Hardware Groups
Multi-Customer Schedule Maintenance
DDOS Mitigation
Includes all Managed Hosting tickets
Billing Tickets
© 2015 IBM Corporation
Cloud IBM SoftLayer
Thank You!
© 2015 IBM Corporation
© 2015 IBM Corporation