Case Studies

Global Hosting Operations

A Fortune 100 Telecommunications and hosting services company manages the infrastructure and IT operations of the clients. They count several fortune 100 companies in Media and Financial sectors as their customers. While they have 70% of their customers in Americas, 100% of their operations are run from their HQ in North America. Also, they are planning to launch offerings in some of the emerging markets in Asia Pac region

Due to the location of the Ops team, teams are required to work night shifts to support other time zones. Not a happy situation for an experienced professional. Also, supporting the Asia Pac customers through US resources is not financially viable. At the same time, the company is worried about the risk to IP if they go with the outsourcing vendors. Also, they have heard horror stories of really junior resources brought by the outsourcing providers.

They have looked at the option of opening a captive centre in India. That is going to be a significant overhead on their management. Though they would want to do that at some point, they decided not to do that now in view of other strategic activities.

The client asked Andor team to suggest a solution that solves their current problem of night shift, makeing their offerings financially viable and improve the customer satisfaction.

For Andor team, this is a typical problem with a twist. For a problem like this, most companies choose to go with a vendor and build a team in locations like India or China. However, this customer feels that outsourcing providers are their competitors and so do not want to expose their IP to them. Also, being premium service providers, they would want top-class talent.

In such a situation, typical answer is to build a captive delivery centre where they could own the IP and can decide on the specific talent. But, they are not in a position to open a captive centre for the following 24 months or so.

So, they really need the best of outsourcing and captive delivery centres. There is an option called ‘Build-Operate-Transfer’ that neatly fits into this requirement. Most of the big and medium players do not want to do this as this requires lot of customized attention and process development as opposed to executing the existing processes for outsourcing deals.

Andor Tech proposed a BOT model for the customer with the customer having an option to transfer the team to their own captive centre at the end of the 24 month period. The customer agreed to the proposal and asked Andor Tech to execute the same.

Andor Tech built a global delivery centre for this customer in Bangalore according to the specific needs of this customer. The centre has redundant power supply, network access, biometric security, 24x7 redundant physical security. Customer’s network is extended to this centre and their global customers are supported from this customer.

Andor has built top-notch team with an average experience of 8 years. 70% of this team left the companies like Microsoft, Oracle, IBM, HP and Cisco to join. In the words of the company CEO when he visited the centre, “You got an awesome team here. I wish we have this kind of team back in HQ”. It is not easy to get that kind of compliment.

Andor has deployed a simple and effective process that governed the entire BOT from requirements, interview process, evaluation, offer making, on-boarding, training and orientation, onsite travel, goal setting and performance review. From the original need of 21 members when they really wanted to test this model, the team grew to over 120 members. This in itself is a testimonial to the success of this project and model.

The team covered different aspects of the operations including, VPDC administration, database administration, Unix/Linux administration, windows administration, middleware administration, network operations centre and service desk.

Eventually, 24 months later, the customer has setup the local captive centre as was originally planned. All of the BOT team members are transferred to the captive centre. The customer is absolutely thrilled with the entire project where they have de-risked themselves at the outset and yet got all the benefits of the captive centre. In fact, they got better as they were able to leverage the experience of Andor team in building such centres earlier.

Cloud Provisioning System

A Fortune 100 IT managed services company provides the on-demand infrastructure and hosting services to their customers. Their in-house infrastructure provisioning team has a 2 month long queue of customer orders on an average. This is affecting the customer sat, order booking and hence revenue recognition. The head of Infra delivery team selected Andor Tech team to study the problem, suggest the solutions and deploy them on approval.

Andor consultants went about this in four steps:

Study

Understand the nature of the problem & why it has been persisting. In the process, understand and document the processes involved in servicing the customer's order.

Analyze

• Identify the processes that are in critical path now & will be in critical paths at different stages until the delivery metrics reach the target level
• Re-engineer each process to:
• Dramatically shorten the time.
• Mitigate the risks non-compliance of the process.
• Instituting automation as and when possible.
• Introduce redundant process/ resources as needed.
• Interlinking the processes through strong communication and escalation matrix.
• Linking up the authority and responsibility in the organization.
• Propose:

Present the proposed process to the customer.

• Deploy

Deploy the new process by leveraging the capabilities of Andor Tech delivery team as needed.

Study

Andor team went about the process of conducting online surveys and reviewing and analysing the data to understand the complete nature of the problem. Some of the findings are:

• There is indeed an average time of 54 days between the request to stand up the infrastructure and the actual delivery of the infrastructure. This has been so for a long time despite few re-organizations and stated intent to improve the process
• Average actual time take to deploy the company software platforms, add into the network matrix is anywhere between 36 hours to 72 hours
• Often the hardware for the kind of the infrastructure ordered is not in stock and had to be ordered
• The need for standing up the order is conveyed by the sales rep once the customer agrees. But the formal request is created in the ticketing system only after the customer completes the documentation process and the system records are created for the customer. Only after the customer record is in the system, a ticket requesting infrastructure for that customer is created. On an average it takes 2-3 weeks for the customer record to be created.
• All the tickets are served by the delivery team on first-come-first-serv principle. No prioritization based on the customer, except for few large customers
• There is a common delivery team that works on all the orders. No specialization based on technologies or services. Consequently, lots of repetition and very limited automation.
• The offerings and customers are global. The provisioning team is situated only in US with limited presence in UK. Consequently, customers in Asia Pac will have longer wait time. Also, email clarifications tend to take days due to time zone difference.

Analyze

The team started by critically examining each of the process, sequence and stated pre-requisites.

• For a multi-billion dollar company offering on-demand infrastructure, why is the hardware not available in inventory?
• It sounds difficult to believe but true. This seems to be a result of couple of things:
• The company is offering very broad types of infrastructure to its customers with the intent on expanding the foot print and driving the revenue.
• Consequently, they became a job shop as opposed to assembly line. To cut the costs of inventory, company had earlier decided to place the order only after customer places the order. Only for the standard items, some levels of inventory are maintained.

This aspect is touching upon the company offerings to it’s customers which is much more strategic than the mandate offered to the Andor team. So, the team decided to focus on optimizing the processes within the constrained environment.

Proposal

Create a separate list of standard and customized services in the offerings. For all the standard services, ensure that the inventory for the next several customers are in place and that there is a regular replenish process. For the customized services, set the expectations to the customers as part of the SLA. Consequently, service costs must reflect these.

Why is the delivery team waiting for the formal creation of the customer record in the system rather than placing the hardware order?

This seems to have been result of a very tight financial process in place at times over riding the operational flexibility. Apparently, some customers who had “committed” did not end up being the customers. Consequently, the “financial” decision was made to ensure that equipment is ordered only after the customer record is created.

While this decision is financially sound in the specific context, it is creating a customer satisfaction issue and is making the very offering “on-demand” questionable.

Proposal

• Since the company has long term relationship with OEMs, it should leverage it to return the equipment if the customer order does not materialize.
• The order for the equipment should be placed once the sales rep makes the request. Each of the sales rep be made “interim customer” in the system for the purpose of placing this order.
• Once the customer formally signs up, the system records be appropriately updated and only then the software installation and deployment of network services happen.
• To mitigate the risk of false orders, the sales rep must be held accountable for the number of such orders annually.

Why is the deployment of software platform on the hardware manual? Why does it take 36-72 hours ? Why are there no specialized teams ? Why is this team located on geographic location ?

The stated answers are

“The company is focussed on the enterprise market as opposed to consumer market. So, the offering is very complex and deployment of that cannot be automated”

“We have so many different offerings. Automation would mean so many different automation threads. Maintaining of that is going to be more difficult”

“We do not really restrict the customers to a particular version of the software out there. They can pretty much choose any version that they want. So, we cannot pre-configure a particular platform for the automation to happen”

All the statements are true in principle and this again points to the company offerings which we are keeping as fixed element in this analysis.

Still, the team sees this as a classical 80-20 situation: 80% of the work or elements of the work is standard and can be fully automated. 20% that varies from order to order will be manual. In reality, though not admitted, the team does not have a band width and direction to build the automation platform. Since they are fighting the fires for being slow (54 days time), they are not in a position to invest on the important element of building automated provisioning system.

Global specialized teams Vs Regional teams: A very important questions. Obviously, we can’t have both. The answer largely depends on the context. In the context of largely manual process, it makes sense to have specialized on the technologies so that we can get the productivity improvement. So, that makes sense for this company at the moment. However, when the company have an automated provisioning system, it really makes sense to have regional teams supported by a centrally located excellence team.

Proposal

• Mandate a small team to build, test and deploy an automated provisioning system. Every order that comes in will be broken into manual and automated components (this is manual step). Once the pre-req manual steps are complete, the automated step should be initiated.
• Depending on the time frame for the above, the company can have a global specialization team followed by the regional support teams.
• With these in place, the team expects that average provisioning time should come down to 8-24 hours depending on the offering.

Proposal

The above proposals are made to the head of the provisioning groups. They are largely accepted. Andor is now mandated to deploy these proposals without touching upon the larger questions of company offerings. Given the constraints on the provisioning team, Andor was asked to build the automated provisioning system and deploy it along with the other proposals. Also, Andor was asked to provide a regional provisioning team to support the Asia Pac business needs.

Deployment

• Andor put together a team of very experienced professionals to build the Automated Provisioning System. The team completed the work in 6 months time. After 8 weeks of testing and fixing, the system went live in the seventh month. Today, 70% of all the standard orders for the company are provisioned through this tool.
• Andor team is asked to build the next version to support the “Build” and “maintenance” of the “Gold” images covering 90% of the company offerings.
• Andor team deployed an Asia Pac provisioning team at their Bangalore delivery centre.

Results

The proposal to pre-order the hardware for the standard offering is still in the deployment level.

Average provisioning time came down from 54 days to 17 days. For the standard offerings, the average provisioning time came down to 32 hours.

ADAS Platform

Product Overview / Key Features

• ADAS Platform based on TI’s TDAx Processor
• LVDS (3 on board and 4 on expansion) and CVBS Cameras (4 on board)
• Support for connecting RADAR and other sensors for the ADAS applications
• Expansion Slot
• LVDS Display, HDMI
• USB 3.0, I2C, GPIOs, SPI, UART
• MIPI CSI-2
• Vision Accelerators to run the ADAS algorithms

Target Applications

• Surround Vision
• Parking Assistance
• Driver Assistance (LDWS, FCWS, Object Detection)
• Adaptive Driving Beam
• ADAS Systems (Front Camera, Rear Camera, Driver Monitoring, TSR detection)
• Camera Monitoring (Mirror Replacement)
• AGVs
• Automotive In-vehicle Telematics

Automotive grade LVDS camera

Product Overview / Key Features

Sensor

• AEC-Q100 qualified High Dynamic Range sensor with full set of automatic controls and image processing pipeline
• 4.2 micron Omnipixel3-HS Pixel technology with best in-class low-light sensitivity of to capture detail-rich color video
• Dynamic range of 115dB
• Sensitivity : 3650 mV/lux-sec
• Pixel Size : 4.2um x 4.2um
• Output format : 10-bit RAW, 8/10-bit YUV 422

Board

• Small form factor board (29mm x 29mm) with microphone for Video and Audio capture
• Onboard ISP for Distortion and Perspective correction at the source side.
• Customizable Lens options (upto 180 deg)
• Serializer supporting FPD Link III technology for LVDS signalling
• Data transmission through COAX or STP cable upto 15 meters
• Power supply, Data, Control signals through the same cable

Multi-Channel Connected Vehicle DVR

Product Overview / Key Features

• Multi-Channel Connected Digital Video Recorder with LVDS & CVBS cameras
• Connects with Head Unit, Telematics Unit, other ECUs thru CAN bus
• Recording with Vehicle Parameters (Speed, Gear Pos, GPS data etc.) to eMMC
• Live camera streaming & Recorded File Streaming to Head Unit via USB and BYOD through Wi-Fi
• Cloud upload for Emergency recorded videos through Telematics
• Various Recording modes
• Automatic Recording (Ignition On)
• Theft mode recording (Triggered by BCM ECU)
• Event based recording (Gsensor)
• User activated recording (Hardwired Switch)
• OTA firmware upgrade via Wi-Fi

Solution for OEM Customers

Being Customized as ECU for Indian OEM

CAN Stack

Product Overview / Key Features

• Modularized CAN Stack
• Easy to configure & adapt to different communication matrix
• Layered Architecture supporting ease of integration
• Generic functions to handle Encoding & Decoding of CAN signals

Features

• CAN 2.0B Active
• Support for CAN-FD
• Support for multiple ports ( e.g. 3 ports)
• Capable to process 100s of PGNs @ 5ms
• Automatic Fault management and recovery mechanisms

Communication Layer

• DLC error
• Message timeout
• Message rolling counter
• Message CRC check
• Signal monitoring

Physical Layer Compliance

• ISO-11898-1 / ISO-11898-2 / ISO-16845

Network Management

• Network Mode management (Reduced CAN and Normal CAN modes)
• Wakeup and sleep on NM message
• CAN bus error handling and bus off recovery
• Network Failures and DTC handling

Datalink Layer Configs

• Baudrate configuration
• Timing accuracy

Intelligent Transport Systems

UBS II based Intelligent Transport Systems for Indian OEM with the following feature

• Main Board with ADAS capability
• Telematics board (4G / GPS)
• Public Information Systems board with 4 displays, Passenger Addressing System & eCALL PoE Board
• IP cameras
• Power Board

Key Features

• Connectivity
• IVN over CAN, Wi-Fi, Ethernet, BT
• DVR Functions
• Support for Analog/IP cameras
• Live video streaming
• Single & Quad video display on BDC
• File sharing over USB and Cloud
• System Diagnostic
• Automatic & On-demand system status
• Embedded browser application
• Sensors
• G-Sensor, Gyrometer, Temp and LDR
• Tracking
• AIS140 compliance
• Vehicle Health Monitoring
• Predective maintanence
• Capability to support
• Automatic passenger counting
• Ticketing Machine
• Digital Techograph
• ADAS Features

In School Surveillance Solution

Product Overview / Key Features

• 4G Streaming camera with Built-in Video recording / storage
• Recorded data pushed to Cloud server.
• Reliable dual storage system (both in Cloud and camera) ensures enhanced security without footage loss.
• Smart Power module in the camera handles the 12v and 24v supply from vehicle battery and protects from cranking and load dump.
• Live Streaming to Cloud Server
• Various Recording modes supported
• Continuous Recording (Ignition On)
• Event based recording (Motion Sensor)
• User Triggered recording (from network)
• Resolution from HD to VGA (customizable) with night Vision IR support
• H.264 encoded recording with configurable bit rate
• Cyclic Overwriting of the record