What’s wrong with our data centre pipes…?

What’s wrong with our data centre pipes…?

Introduction…

The WAN market has seen some dramatic changes over the last few years, we now see organisations moving from Mb WAN link speeds to Gb WAN speeds. At the same time, everyone is experiencing the need to move very large volumes of data.

This brings a whole new raft of problems to the table, everyone is trying to maximise their data transport layers to return the investment they have made in their WAN infrastructure.

There is no one silver bullet that covers all aspects of accelerating data across a WAN – each has its strength and weaknesses, even we shy away if we are not a good fit or we are unable to accelerate your data over the WAN.

What do we do…

We provide WAN acceleration technology to customers who need to move data over distance, that distance we need to move data can be considered as regional within the UK, regional to the cloud, regional to EMEA or regional to transatlantic data centres.

What type of data volumes will we accelerate, listed below are few examples:

  • Backup & Recovery
  • Storage Replication
  • Data Migration
  • WANdisco
  • Big Data
  • Global Workflows
  • Amazon AWS Cloud Data
  • Object S3 Data

What is the challenge…

A good example is a 1Gb/s WAN link with a 5ms latency, this will limit your data throughput to 50% or less of the available bandwidth. Double this latency to 10ms and you halve the throughput. Let’s now add packet loss to the equation and your performance takes a nose dive, furthermore add packet re-routing and we are on our knee’s begging for mercy.

Initial reaction from customers is that they “fill their bandwidth already” or “we will just add more bandwidth”, doing so does not help at all. In fact, it just compounds the problem.

We all like pictures and graphs…

Everyone likes pictures and graphs rather than mumbo jumbo technical stuff, we want to depict the effects that are commonly seen but not talked about.

 

Remember, we are not here to pick a fight with your network team. However, they cannot fix application, WAN packet loss or when packets are re-routed across the WAN.

 

The above table is a sample of global ping statistics that we grabbed today, it is all scary when you start charting this type of data.

How on earth do we do this…?

We move data regardless of distance utilising up to 98% of your maximum possible bandwidth. We do NOT touch your data!

We achieve this by creating parallelism within the WAN and managing each virtual link with Artificial Intelligence (AI) to maximise your data throughput, manage the packet loss and congestion. Remember, nor do we NOT manipulate the data itself.

All data is equal…

100% transparent, we will move your data at near wire speed from point A to point B regardless of latency and mitigating packet loss.

 

All data is equal, including encrypted, encoded (video | images | scans | DICOM | PACS etc.) not forgetting deduplicated and compressed data.

No agents or application integration is required, we will not tie you into any proprietary orchestration layers. We will continue to use your existing and preferred methods, e.g. FTP, DFSR, Object etc.

Making industry standard TCP/IP work…

  • Scalable up to 40Gb/s today, with 100Gb WAN support in 2018
  • Multi-protocol (S3, SnapMirror, Commvault, FTP etc.) simultaneously
  • Mitigates chattiness at the application level
  • Fixes your throughput problem
  • Mitigates noisy networks (packet loss)

WAN Link

Link size Latency Example geography
100Mb/s 100ms+ US to Europe, Europe to APAC
1Gb/s 10ms+ as above and inter region UK to Frankfurt, Gibraltar to Sweden
10Gb/s 10ms+ as above plus intra country, e.g. London to Manchester
Multiple 10Gb/s 5ms+ as above plus Inter City metro

Moving data over distance guide

Unaccelerated, s=seconds, m=minutes, h=hours, d=days
Latency
Size Bandwidth Accelerated 10ms 20ms 40ms 100ms 200ms
5Gb 100Mb/s 7.6m 8.3m* 13.7m 34.1m 1.1h
600Mb/s 1.3m 3.4m 6.8m
1Gb/s 45s
10Gb/s 4.6s
20GB 100Mb/s 30.3m 33.3m* 54.6m 2.3h 4.6h
600Mb/s 5.1m 13.7m 27.3m
1Gb/s 3m
10Gb/s 18.2s
30GB 100Mb/s 45.5m 50m* 1.4h 3.5h 6.8h
600Mb/s 7.7m 20.5m 41m
1Gb/s 4.6m
10Gb/s 27.3s
100GB 100Mb/s 2.5h 2.8h* 4.6h 11.4h 22.8h
600Mb/s 25.5m 1.1h 2.3h
1Gb/s 15.2m
10Gb/s 1.5m
2TB 100Mb/s 2.2d 2.4days* 3.9days 9.7days 19.4days
600Mb/s 8.7h 23.3h 1.9days
1Gb/s 5.2h
10Gb/s 31.1m

 

Use Cases…

WAN Acceleration means that you can now move data as efficiently as possible, dramatically reducing the time needed to complete business critical operations therefore achieving a much higher ROI of the WAN link costs and the opportunity to reduce WAN costs.

For several customers, the ROI was almost immediate. The technology pays for itself by offsetting a ‘red herring’ investment or eliminating service costs.

The process of incorporating the technology into an environment is minimal and takes minutes to deploy.  Normally no changes are needed to the underlying infrastructure, applications and workflows.

Use cases are only limited by the functions that a given customer needs to deliver.  Some common example use-cases include;

Backup and Recovery (Major Retailer)

  • Recovery Time Objective (RTO) more than 15 hours estimated (it was closer to 30 hours)
  • This was reduced to less than 5 hours. 430Gb of data over 2,800 miles (600 Mb/s WAN link with 86ms of latency and 1% packet loss)
  • Eliminated Iron Mountain service requirement
  • Additional benefits included no need to upgrade bandwidth
  • Removed audit, compliance and risk
  • Reduced physical tape cartridge usage
  • Cost savings from Iron Mountain alone £260,000 per year

Replication (Multi-National Insurance company)

  • Despite investing heavily in dark fibre between 2 locations less than 2ms apart, to achieve simultaneous cross-site replication the company’s replication window was consistently being missed with a large percentage of the replication attempts ending in failure
  • This had serious implications on Recovery Time Objective (RTO) which in the event of a disaster would have clear consequence for customer experience and therefore revenues
  • Less than a 35 minute to install both appliances
  • The POC was underway with the company’s 2 existing 10Gb backup pipes
  • The units were installed in a bow tie configuration for failover purposes
  • Transfer times were now less than half of those previous
  • The real test was to complete the simultaneous bi-directional replication, this completed without incident and at the accelerated rate
  • RTO was back to where it needed to be to comply with the service level agreement, high on the Disaster Recover/Business Continuity Agenda for Executive Management
  • The dark fibre was no longer required
  • The advice from their hardware vendor had been to replicate the complete dark fibre environment; this investment was no longer required. Just these two elements combined resulted in a cost avoidance of over £435,000 which more than justified the purchase.

Big Data (Major European Bank)

  • The investment in a Big Data warehouse for customer analytics was significant but the value of the investment was being questioned at the executive level as they couldn’t get the data into the warehouse in a timely manner ‘Time-to-Intelligence’
  • Basically, it was taking too long to get the data into the warehouse form their European operations
  • With no changes to their environment and no WAN upgrades across various WAN capacities and distances (latencies)
  • In their case the ROI was instant and unprecedented, the technology had a measurable ROI against a proposed WAN upgrade programme that has now been shelved as it was unnecessary

Global workflow and distribution (Media Company)

  • Distributing advertising for page and TV has always been a challenge for which there appeared to be no answer
  • We enabled the organisation to reduce the time it takes not only to distribute content but enable global collaboration and workflow which has enabled them to be much more efficient and agile as a business essential in today’s fast-moving markets
  • Essential to compete in a market where advertisers want to rollout (and change) marketing programmes globally in hours and days not weeks and months

So, what’s next…?

You all have the option to say, “no thanks”. But remember, we did not create this problem, but we can solve it for you and it WILL NOT cost you your whole year’s budget.

Methods of Contact

Telephone: 0118 324 1300

We will listen, ask a few questions and maybe we might be able to help.

Email: bridgeworks@symstor.com

Web: www.symstor.com/bridgeworks

We have a contact form and a bandwidth calculator to play with.

Events: www.symstor.com/events

Register for one of our upcoming one-many WebEx events where you can view the presentation, demonstration and ask questions.